Skip to content

A TypeScript library to serialize and deserialize object graph from/to JSON in a fast and non-recursive way

License

Notifications You must be signed in to change notification settings

zircon63/json-object-mapper

 
 

Repository files navigation

Introduction

json-object-mapper is a typescript library designed to serialize and de-serialize DTO objects from and to JSON objects. Using the library, you would be able to load the JSON data from Http/File/Stream stright into an object graph of your DTO classes as well as serialize a DTO object graph so that it can be sent to an output stream.

The idea behind this is that you do not need to add serialization and de-serialization methods to each of your DTO classes - thus keeping them clean and simple.

Build Status

Usage

The core of the library is the ObjectMapper class which essentially serializes and de-serializes object graph.

It has the following methods:

  1. serialize(Object) => String : will serialize a class instance into JSON string.
  2. deserialize(Type, Object) => Type : will take a class type and a JSON object and create a new instance of the class type based on the JSON data model.
  3. deserializeArray(Type, Object) => Type[] : will take a JSON array and convert that to an array of class type.

There is also the @JsonProperty decorator which has additional metadata about how the class properties needs to be processed. This decorator comes handy when you do want to deserialize a non-primitive type object to a instance property (see JS primitive types). It's the same when declaring arrays even though the array can be of a primitive date type. For non-primitive types if this property decorator is not present, raw JSON data will be assinged as the property value.

The decorator takes the following optional values:

  • name : name of the JSON property to map to. This name will also be used during serialization
  • required: is this field required in the JSON object that is being deserialized
  • access: is this serializable and de-serializable
  • type: the type of Object that should be assigned to this property

Lets have a look at an example:

class SimpleRoster {
    @JsonProperty()
    private name: String;
    @JsonProperty()
    private worksOnWeekend: Boolean;
    @JsonProperty()
    private numberOfHours: Number;
    @JsonProperty({type:Date})
    private systemDate: Date;

    public isAvailableToday(): Boolean {
        if (this.systemDate.getDay() % 6 == 0 && this.worksOnWeekend == false) {
            return false;
        }
        return true;
    }

}
let json = {
    'name': 'John Doe',
    'worksOnWeekend' : false,
    'numberOfHours': 8,
    'systemDate' : 1483142400000 // Sat Dec 31, 2016
};

let testInstance: SimpleRoster = ObjectMapper.deserialize(SimpleRoster, json);
expect(testInstance.isAvailableToday()).toBeFalsy();

First lets talk about the property systemDate and the @JsonProperty property decorator assinged to it. As the Date type is not a primitive data type, you will need to explicitly mention that to the processor that we need to create a new instance of Date object with the supplied value and assign it to the systemDate property.

When the deserialize method runs, it will first create a new instance of 'SimpleRoster' class. Then it will parse the keys of the instance and assign the correct values to each keys. While doing that, it will make sure that the right property types are maintained - meaning that name field will be assinged a String object with the value 'John Doe', numberOfHours will be assigned a Number object and the systemDate field will be assinged a Date object with the value Sat Dec 31, 2016. The method will also make sure that all the nested object graphs has been created based on the JSON model.

The serialize method will serialize an object graph into JSON string (similar to what JSON.stringrify() method would do) while honoring the decorator metadata. Following is an example of serialization:

class SimpleClass {
    firstName: string = "John";
    middleName: string = "P";
    lastName: string = "Doe";
    @JsonIgnore()
    password: string = "mypwd";
    @JsonProperty({ type: String, name: "AKA" })
    knownAs: String[] = ["John", "Doe", "JohnDoe", "JohnPDoe"]
};


let instance: SimpleClass = new SimpleClass();

let stringrified: String = ObjectMapper.serialize(instance);
expect(stringrified).toBe('{"firstName":"John","middleName":"P","lastName":"Doe","AKA":["John","Doe","JohnDoe","JohnPDoe"]}');

From the example above, the knownAs property is serialized as AKA as defined in the @JsonProperty decorator. The password property does not serialized as defined in the @JsonIgnore decorator.

The library uses non-recursive iterations to process data. That means you can serialize and de-serialize large amount of data quickly and efficiently using any modern browser as well as native applications (such as nodejs, electron, nativescript, etc.). Please have a look at the test folder to see few examples of using large dataset as well as how to use the library in general.

Inspiration

The library is inspired by the popular jackson library. As with the java version, it uses typescript decorators to decorate properties with additional metadata. I will try to stick as close as possible to the jackson annotations so that the transition can be simple.

Installation

The project is hosted on github.com. You can either download it from there or from npmjs.com with the npm command:

npm install json-object-mapper --save

Depedency

The library depends on reflect-metadata library. Originally, I thought of not having to force a dependency on another library. But when you look the geneated JS code for a typescript decorator, it is always trying to check for the variable Reflect. So, until such time when typescript decorators can be de-coupled from Reflect library, I am sticking with it. You can either download reflect-metadata from there or from npmjs.com with the npm command:

npm install reflect-metadata --save

and make sure to import it in a global place, like app.ts:

import "reflect-metadata";

If you are using Angular 2 you should already have this shim installed.

Things to remember

Enum serialization and de-serialization

You can use enum data type by specifying the type property of @JsonProperty decorator. You will need to use Serializer and Deserializer to make the enum work correctly.

Following is an example of enum serialization and de-serialization:

@CacheKey("DaysEnumSerializerDeserializer")
class DaysEnumSerializerDeserializer implements Deserializer, Serializer{
    deserialize = (value: string): Days => {
        return Days[value];
    }
    serialize = (value: Days): string => {
        return '"' + Days[value] + '"';
    }
}

enum Days{
    Sun, Mon, Tues, Wed, Thurs, Fri, Sat
}  

class Workday{
    @JsonProperty({ type: Days, deserializer: DaysEnumSerializerDeserializer, serializer: DaysEnumSerializerDeserializer})
    today: Days;
}        

let json = { "today": 'Tues' };

let testInstance: Workday = ObjectMapper.deserialize(Workday, json);
expect(testInstance.today == Days.Tues).toBeTruthy();
testInstance.today = Days.Fri;
let serialized: String = ObjectMapper.serialize(testInstance);
expect(serialized).toBe('{"today":"Fri"}');

Map objects serialization and de-serialization

This can be achieved by using implementations of Serializer and Deserializer. For example:

class MapDeserializer implements Deserializer {
    deserialize = (value: any): any => {
        let mapToReturn: Map<String, String> = new Map<String, String>();
        if (value) {
            Object.keys(value).forEach((key: String) => {
                mapToReturn.set(key, value['' + key]);
            });
        }
        return mapToReturn;
    }
}

A special thing about Date object

It's very hard to get a Date instance right across all browsers - have a look at the stackoverflow discussion. The best way to manage this complexitiy I have found so far is to use the new Date(value) constructor which takes the number of milliseconds since 1st January 1970 UTC. So, to best use this library, make sure that the date is passed on as the number of milliseconds during deserializition:

jsonTest["dateType"] : 1333065600000;

even though it will take the date as string without guranting the accuracy of time:

jsonTest["dateType"] : '05/08/2013';

For serialization, it will only print out the milliseconds:

{"test":1457096400000}

Also, you will need to use the DateSerializer or your own implementation for serializing Date objects.

@JsonProperty({type: Date, name:'dateOfBirth', serializer: DateSerializer})
dob: Date = new Date(1483142400000)

About

A TypeScript library to serialize and deserialize object graph from/to JSON in a fast and non-recursive way

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • TypeScript 99.9%
  • JavaScript 0.1%