Serialize and deserialize complex object graphs to JSON
Serializr is feature complete, and easily extendable. Since there are no active maintainers the project is frozen feature wise. Bug reports are welcome and will be addressed.
Want to maintain a small open source project or having great ideas for this project? We are looking for maintainers, so apply!
Serializr is a utility library that helps converting json structures into complex object graphs and the other way around. For a quick overview, read the introduction blog post
Features:
Non-features:
From npm: npm install serializr --save
From CDN: https://unpkg.com/serializr which declares the global serializr object.
import {
createModelSchema,
primitive,
reference,
list,
object,
identifier,
serialize,
deserialize,
} from 'serializr';
// Example model classes
class User {
uuid = Math.floor(Math.random() * 10000);
displayName = 'John Doe';
}
class Message {
message = 'Test';
author = null;
comments = [];
}
function fetchUserSomewhere(uuid) {
// Lets pretend to actually fetch a user; but not.
// In a real app this might be a database query
const user = new User();
user.uuid = uuid;
user.displayName = `John Doe ${uuid}`;
return user;
}
function findUserById(uuid, callback, context) {
// This is a lookup function
// identifier is the identifier being resolved
// callback is a node style calblack function to be invoked with the found object (as second arg) or an error (first arg)
// context is an object detailing the execution context of the serializer now
callback(null, fetchUserSomewhere(uuid));
}
// Create model schemas
createModelSchema(Message, {
message: primitive(),
author: reference(User, findUserById),
comments: list(object(Message)),
});
createModelSchema(User, {
uuid: identifier(),
displayName: primitive(),
});
// can now deserialize and serialize!
const message = deserialize(Message, {
message: 'Hello world',
author: 17,
comments: [
{
message: 'Welcome!',
author: 23,
},
],
});
const json = serialize(message);
console.dir(message, { colors: true, depth: 10 });
With decorators (TypeScript or ESNext) building model schemas is even more trivial:
import {
createModelSchema,
primitive,
reference,
list,
object,
identifier,
serialize,
deserialize,
getDefaultModelSchema,
serializable,
} from 'serializr';
class User {
@serializable(identifier())
uuid = Math.random();
@serializable displayName = 'John Doe';
}
class Message {
@serializable message = 'Test';
@serializable(object(User))
author = null;
// Self referencing decorators work in Babel 5.x and Typescript. See below for more.
@serializable(list(object(Message)))
comments = [];
}
// You can now deserialize and serialize!
const message = deserialize(Message, {
message: 'Hello world',
author: { uuid: 1, displayName: 'Alice' },
comments: [
{
message: 'Welcome!',
author: { uuid: 1, displayName: 'Bob' },
},
],
});
console.dir(message, { colors: true, depth: 10 });
// We can call serialize without the first argument here
//because the schema can be inferred from the decorated classes
const json = serialize(message);
Decorator: Caveats
Babel 6.x does not allow decorators to self-reference during their creation, so the above code would not work for the Message class. Instead write:
class Message {
@serializable message = 'Test';
@serializable(object(User))
author = null;
comments = [];
constructor() {
getDefaultModelSchema(Message).props['comments'] = list(
object(Message)
);
}
}
TypeScript
Enable the compiler option experimentalDecorators in tsconfig.json or pass it as flag --experimentalDecorators to the compiler.
Babel 5.x
{
"stage": 1
}
Babel 6.x:
Install support for decorators: npm i --save-dev babel-plugin-transform-decorators-legacy. And enable it in your .babelrc file:
{
"presets": ["es2015", "stage-1"],
"plugins": ["transform-decorators-legacy"]
}
Probably you have more plugins and presets in your .babelrc already, note that the order is important and transform-decorators-legacy should come as first.
The two most important functions exposed by serializr are serialize(modelschema?, object) -> json tree and deserialize(modelschema, json tree) -> object graph.
What are those model schemas?
The driving concept behind (de)serialization is a ModelSchema. It describes how model object instances can be (de)serialize to json.
A simple model schema looks like this:
const todoSchema = {
factory: context => new Todo(),
extends: ModelSchema,
props: {
modelfield: PropSchema,
},
};
The factory tells how to construct new instances during deserialization.
The optional extends property denotes that this model schema inherits its props from another model schema.
The props section describes how individual model properties are to be (de)serialized. Their names match the model field names.
The combination fieldname: true is simply a shorthand for fieldname: primitive()
For convenience, model schemas can be stored on the constructor function of a class. This allows you to pass in a class reference wherever a model schema is required. See the examples below.
PropSchemas contain the strategy on how individual fields should be serialized. It denotes whether a field is a primitive, list, whether it needs to be aliased, refers to other model objects etc. PropSchemas are composable. See the API section below for the details, but these are the built-in property schemas:
primitive(): Serialize a field as primitive valueidentifier(): Serialize a field as primitive value, use it as identifier when serializing references (see reference)date(): Serializes dates (as epoch number)alias(name, propSchema): Serializes a field under a different namelist(propSchema): Serializes an array based collectionmap(propSchema): Serializes an Map or string key based collectionobject(modelSchema): Serializes an child model elementreference(modelSchema, lookupFunction?): Serializes a reference to another model elementcustom(serializeFunction, deserializeFunction): Create your own property serializer by providing two functions, one that converts modelValue to jsonValue, and one that does the inverse"*": true that serializes all enumerable, non mentioned values as primitiveIt is possible to define your own prop schemas. You can define your own propSchema by creating a function that returns an object with the following signature:
{
serializer: (sourcePropertyValue: any) => jsonValue,
deserializer: (jsonValue: any, callback: (err, targetPropertyValue: any) => void, context?, currentPropertyValue?) => void
}
For inspiration, take a look at the source code of the existing ones on how they work, it is pretty straightforward.
The context object is an advanced feature and can be used to obtain additional context-related information about the deserialization process.
context is available as:
ref prop schema's (see below)deserializer of a custom propSchemaWhen deserializing a model elememt / property, the following fields are available on the context object:
json: Returns the complete current json object that is being deserializedtarget: The object currently being deserialized. This is the object that is returned from the factory function.parentContext: Returns the parent context of the current context. For example if a child element is being deserialized, the context.target refers to the current model object, and context.parentContext.target refers to the parent model object that owns the current model object.args: If custom arguments were passed to the deserialize / update function, they are available as context.args.JSDOC type defintions for usage w/o typescript.
Type: object
Parameters
propsvalue anywriteable booleanget (Function | undefined)set (Function | undefined)configurable booleanenumerable booleansourcePropertyValue anyjsonValue anycallback cpsCallbackcontext ContextcurrentPropertyValue anyid anytarget objectcontext Contextresult anyerror anyid stringcallback cpsCallbackfactorytargetClassProperties
serializer serializerFunctiondeserializer deserializerFunctionidentifier booleanReturns any any - serialized object
Returns any void
Returns any void
Returns any void
Creates a model schema that (de)serializes from / to plain javascript objects.
Its factory method is: () => ({})
Parameters
props object property mapping,Examples
var todoSchema = createSimpleSchema({
title: true,
done: true,
});
var json = serialize(todoSchema, { title: 'Test', done: false });
var todo = deserialize(todoSchema, json);
Returns object model schema
Creates a model schema that (de)serializes an object created by a constructor function (class).
The created model schema is associated by the targeted type as default model schema, see setDefaultModelSchema.
Its factory method is () => new clazz() (unless overriden, see third arg).
Parameters
clazz (constructor | class) class or constructor functionprops object property mappingfactory function optional custom factory. Receives context as first argExamples
function Todo(title, done) {
this.title = title;
this.done = done;
}
createModelSchema(Todo, {
title: true,
done: true,
});
var json = serialize(new Todo('Test', false));
var todo = deserialize(Todo, json);
Returns object model schema
Decorator that defines a new property mapping on the default model schema for the class it is used in.
When using typescript, the decorator can also be used on fields declared as constructor arguments (using the private / protected / public keywords).
The default factory will then invoke the constructor with the correct arguments as well.
Parameters
arg1arg2arg3Examples
class Todo {
Returns PropertyDescriptor
The serializeAll decorator can be used on a class to signal that all primitive properties should be serialized automatically.
Parameters
targetReturns the standard model schema associated with a class / constructor function
Parameters
thing objectReturns ModelSchema model schema
Sets the default model schema for class / constructor function.
Everywhere where a model schema is required as argument, this class / constructor function
can be passed in as well (for example when using object or ref.
When passing an instance of this class to serialize, it is not required to pass the model schema
as first argument anymore, because the default schema will be inferred from the instance type.
Parameters
clazz (constructor | class) class or constructor functionmodelSchema ModelSchema a model schemaReturns ModelSchema model schema
Serializes an object (graph) into json using the provided model schema. The model schema can be omitted if the object type has a default model schema associated with it. If a list of objects is provided, they should have an uniform type.
Parameters
arg1 modelschema to use. Optionalarg2 object(s) to serializeReturns object serialized representation of the object
Deserializes a json structor into an object graph. This process might be asynchronous (for example if there are references with an asynchronous lookup function). The function returns an object (or array of objects), but the returned object might be incomplete until the callback has fired as well (which might happen immediately)
Parameters
schema (object | array) to use for deserializationjson json data to deserializecallback function node style callback that is invoked once the deserializaiton has finished.
First argument is the optional error, second argument is the deserialized object (same as the return value)customArgs any custom arguments that are available as context.args during the deserialization process. This can be used as dependency injection mechanism to pass in, for example, stores.Returns (object | array) deserialized object, possibly incomplete.
Similar to deserialize, but updates an existing object instance. Properties will always updated entirely, but properties not present in the json will be kept as is. Further this method behaves similar to deserialize.
Parameters
modelSchema object , optional if it can be inferred from the instance typetarget object target instance to updatejson object the json to deserializecallback function the callback to invoke once deserialization has completed.customArgs any custom arguments that are available as context.args during the deserialization process. This can be used as dependency injection mechanism to pass in, for example, stores.Indicates that this field contains a primitive value (or Date) which should be serialized literally to json.
Examples
createModelSchema(Todo, {
title: primitive(),
});
console.dir(serialize(new Todo('test')));
// outputs: { title : "test" }
Returns ModelSchema
Similar to primitive, but this field will be marked as the identifier for the given Model type.
This is used by for example reference() to serialize the reference
Identifier accepts an optional registerFn with the signature:
(id, target, context) => void
that can be used to register this object in some store. note that not all fields of this object might
have been deserialized yet.
Parameters
registerFn RegisterFunction optional function to register this object during creation.Examples
var todos = {};
var s = _.createSimpleSchema({
id: _.identifier((id, object) => (todos[id] = object)),
title: true,
});
_.deserialize(s, {
id: 1,
title: 'test0',
});
_.deserialize(s, [{ id: 2, title: 'test2' }, { id: 1, title: 'test1' }]);
t.deepEqual(todos, {
1: { id: 1, title: 'test1' },
2: { id: 2, title: 'test2' },
});
Returns PropSchema
Similar to primitive, serializes instances of Date objects
Alias indicates that this model property should be named differently in the generated json. Alias should be the outermost propschema.
Parameters
name string name of the json field to be used for this propertypropSchema PropSchema propSchema to (de)serialize the contents of this fieldExamples
createModelSchema(Todo, {
title: alias('task', primitive()),
});
console.dir(serialize(new Todo('test')));
// { task : "test" }
Returns PropSchema
Can be used to create simple custom propSchema.
Parameters
serializer function function that takes a model value and turns it into a json valuedeserializer function function that takes a json value and turns it into a model valueExamples
var schema = _.createSimpleSchema({
a: _.custom(
function(v) {
return v + 2;
},
function(v) {
return v - 2;
}
),
});
t.deepEqual(_.serialize(s, { a: 4 }), { a: 6 });
t.deepEqual(_.deserialize(s, { a: 6 }), { a: 4 });
Returns PropSchema
object indicates that this property contains an object that needs to be (de)serialized
using its own model schema.
N.B. mind issues with circular dependencies when importing model schema's from other files! The module resolve algorithm might expose classes before createModelSchema is executed for the target class.
Parameters
modelSchema ModelSchema to be used to (de)serialize the objectExamples
class SubTask {}
class Todo {}
createModelSchema(SubTask, {
title: true,
});
createModelSchema(Todo, {
title: true,
subTask: object(SubTask),
});
const todo = deserialize(Todo, {
title: 'Task',
subTask: {
title: 'Sub task',
},
});
Returns PropSchema
reference can be used to (de)serialize references that point to other models.
The first parameter should be either a ModelSchema that has an identifier() property (see identifier)
or a string that represents which attribute in the target object represents the identifier of the object.
The second parameter is a lookup function that is invoked during deserialization to resolve an identifier to an object. Its signature should be as follows:
lookupFunction(identifier, callback, context) where:
1. identifier is the identifier being resolved
2. callback is a node style calblack function to be invoked with the found object (as second arg) or an error (first arg)
3. context see context.
The lookupFunction is optional. If it is not provided, it will try to find an object of the expected type and required identifier within the same JSON document
N.B. mind issues with circular dependencies when importing model schemas from other files! The module resolve algorithm might expose classes before createModelSchema is executed for the target class.
Parameters
target : ModelSchema or stringlookupFn RefLookupFunction functionExamples
class User {}
class Post {}
createModelSchema(User, {
uuid: identifier(),
displayname: primitive(),
});
createModelSchema(Post, {
author: reference(User, findUserById),
message: primitive(),
});
function findUserById(uuid, callback) {
fetch('http://host/user/' + uuid)
.then(userData => {
deserialize(User, userData, callback);
})
.catch(callback);
}
deserialize(
Post,
{
message: 'Hello World',
author: 234,
},
(err, post) => {
console.log(post);
}
);
Returns PropSchema
List indicates that this property contains a list of things. Accepts a sub model schema to serialize the contents
Parameters
propSchema PropSchema to be used to (de)serialize the contents of the arrayExamples
class SubTask {}
class Task {}
class Todo {}
createModelSchema(SubTask, {
title: true,
});
createModelSchema(Todo, {
title: true,
subTask: list(object(SubTask)),
});
const todo = deserialize(Todo, {
title: 'Task',
subTask: [
{
title: 'Sub task 1',
},
],
});
Returns PropSchema
Similar to list, but map represents a string keyed dynamic collection. This can be both plain objects (default) or ES6 Map like structures. This will be inferred from the initial value of the targetted attribute.
Parameters
propSchema anySimilar to map, mapAsArray can be used to serialize a map-like collection where the key is contained in the 'value object'.
Example: consider Map<id: number, customer: Customer> where the Customer object has the id stored on itself.
mapAsArray stores all values from the map into an array which is serialized.
Deserialization returns a ES6 Map or plain object object where the keyPropertyName of each object is used for keys.
For ES6 maps this has the benefit of being allowed to have non-string keys in the map. The serialized json also may be slightly more compact.
Parameters
propSchema any , {string} keyPropertyName - the property of stored objects used as key in the mapkeyPropertyNameconst todoSchema = {
factory: () => {},
props: {
task: primitive(),
owner: reference('_userId', UserStore.findUserById), // attribute of the owner attribute of a todo + lookup function
subTasks: alias('children', list(object(todoSchema))),
},
};
const todo = deserialize(
todoSchema,
{ task: 'grab coffee', owner: 17, children: [] },
(err, todo) => {
console.log('finished loading todos');
}
);
const todoJson = serialize(todoSchema, todo);
function Todo(parentTodo) {
this.parent = parentTodo; // available in subTasks
}
const todoSchema = {
factory: context => new Todo(context.parent),
props: {
task: primitive(),
owner: reference('_userId', UserStore.findUserById), // attribute of the owner attribute of a todo + lookup function
subTasks: alias('children', list(object(todoSchema))),
},
};
setDefaultModelSchema(Todo, todoSchema);
const todo = deserialize(
Todo, // just pass the constructor name, schema will be picked up
{ task: 'grab coffee', owner: 17, children: [] },
(err, todos) => {
console.log('finished loading todos');
}
);
const todoJson = serialize(todo); // no need to pass schema explicitly
function Todo() {}
// creates a default factory, () => new Todo(), stores the schema as default model schema
createModelSchema(Todo, {
task: primitive(),
});
const todo = deserialize(
Todo, // just pass the constructor name, schema will be picked up
{ task: 'grab coffee', owner: 17, children: [] },
(err, todos) => {
console.log('finished loading todos');
}
);
const todoJson = serialize(todo); // no need to pass schema explicitly
class Todo {
@serializable(primitive())
task = 'Grab coffee';
@serializable(reference('_userId', UserStore.findUserById))
owner = null;
@serializable(alias('children', list(object(todoSchema))))
subTasks = [];
}
// note that (de)serialize also accepts lists
const todos = deserialize(
Todo,
[
{
task: 'grab coffee',
owner: 17,
children: [],
},
],
(err, todos) => {
console.log('finished loading todos');
}
);
const todoJson = serialize(todos);
const someTodoStoreById = {};
getDefaultModelSchema(Todo).factory = context => {
const json = context.json;
if (someTodoStoreById[json.id]) return someTodoStoreById[json.id]; // reuse instance
return (someTodoStoreById[json.id] = new Todo());
};
This pattern is useful to avoid singletons but allow to pass context specific data to constructors. This can be done by passing custom data to deserialize / update as last argument,
which will be available as context.args on all places where context is available:
class User {
constructor(someStore) {
// User needs access to someStore, for whatever reason
}
}
// create model schema with custom factory
createModelSchema(User, { username: true }, context => {
return new User(context.args.someStore);
});
// don't want singletons!
const someStore = new SomeStore();
// provide somestore through context of the deserialization process
const user = deserialize(
User,
someJson,
(err, user) => {
console.log('done');
},
{
someStore: someStore,
}
);
// models.js:
import { observable, computed } from 'mobx';
import { serializable, identifier } from 'serializr';
function randomId() {
return Math.floor(Math.random() * 100000);
}
export class Box {
<<<<<<< HEAD
@serializable(identifier())
id = randomId();
@serializable
@observable
x = 0;
@serializable
@observable
y = 0;
@serializable
@observable
location = 0;
constructor(location, x, y) {
=======
@serializable(identifier()) id = randomId();
@serializable @observable x = 0;
@serializable @observable y = 0;
@serializable @observable location = 0;
constructor(location, x, y){
>>>>>>> Updated readme and changelog
this.location = location;
this.x = x;
this.y = y;
}
<<<<<<< HEAD
@serializable
@computed
get area() {
=======
@serializable @computed get area() {
>>>>>>> Updated readme and changelog
return this.x * this.y;
}
}
export class Arrow {
@serializable(identifier())
id = randomId();
@serializable(reference(Box))
from;
@serializable(reference(Box))
to;
}
// store.js:
import { observable, transaction } from 'mobx';
import {
createSimpleSchema,
identifier,
list,
serialize,
deserialize,
update,
} from 'serializr';
import { Box, Arrow } from './models';
// The store that holds our domain: boxes and arrows
const store = observable({
boxes: [],
arrows: [],
selection: null,
});
// Model of the store itself
const storeModel = createSimpleSchema({
boxes: list(object(Box)),
arrows: list(object(Arrow)),
selection: reference(Box),
});
// Example Data
// You can push data in as a class
store.boxes.push(new Box('Rotterdam', 100, 100), new Box('Vienna', 650, 300));
// Or it can be an raw javascript object with the right properties
store.arrows.push({
id: randomId(),
from: store.boxes[0],
to: store.boxes[1],
});
// (de) serialize functions
function serializeState(store) {
return serialize(storeModel, store);
}
function deserializeState(store, json) {
transaction(() => {
update(storeModel, store, json);
});
}
// Print ... out for debugging
console.dir(serializeState(store), { depth: 10, colors: true });
"*": true respect extends clauses1.1.12
class constructor cannot be invoked without 'new' error, see #42 by @pinksquidhatmapAsArray, to support maps that serialize to arrays with objects with identifiers. And vice versa. By @vonovak. See #22FAQs
Serialize and deserialize complex object graphs to JSON
The npm package serializr receives a total of 14,791 weekly downloads. As such, serializr popularity was classified as popular.
We found that serializr demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 5 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Research
Socket researchers have discovered malicious npm packages targeting crypto developers, stealing credentials and wallet data using spyware delivered through typosquats of popular cryptographic libraries.
Security News
Socket's package search now displays weekly downloads for npm packages, helping developers quickly assess popularity and make more informed decisions.
Security News
A Stanford study reveals 9.5% of engineers contribute almost nothing, costing tech $90B annually, with remote work fueling the rise of "ghost engineers."