fashion-model

JavaScript library for defining types and their properties with support for wrapping/unwrapping, serialization/deserialization, validation, and JSON schema.

Installation

npm install fashion-model --save

Overview

The fashion-model module provides utility code for defining data model types. These data model types provide helpful accessor methods (getters and setters) for the properties defined for the model type. This module is compatible with both web browser and Node.js runtime environments.

These models can be thought of as a "schema" that provide extra safeguards for working with objects. These model types are not tied to a specific data storage backend so you can use these in the browser or on the server-side with very little overhead. This approach to defining your schema is similar to Mongoose schemas except that this library is not tied to MongoDB or any other storage engine.

If you application is fetching data from the database for a client request and there is no need to process the data, then simply serialize the data without creating Model instances to wrap the data. Creating Model instances creates unnecessary overhead with no benefit (this is the default behavior of Mongoose). However, if you're accessing or setting properties on an object then you might find it helpful to wrap the raw object with a Model instance and use the getters and setters to work with the data.

Relationship to JSON Schema

Model definitions are similar to JSON Schema and, when possible, similar naming conventions were chosen. However, this module is more tailored to runtime usage. If desired, you can convert your model definitions to a JSON Schema representation fairly easily. See JSON Schema section for more information.

Usage

Requiring

// Requiring the base Model type
const Model = require('fashion-model');

// Variation on requiring the Model type
const Model = require('fashion-model/Model');

// Requiring the Enum type factory
const Enum = require('fashion-model/Enum');

// Create new model type
const NewModel = Model.extend(config);

// Create new enum type
const NewEnum = Enum.create(config);

Primitive Types

The following primitive types are supported:

Data Type	JavaScript Type	Alias	Model Type
Any		"any"	require("fashion-model/Any")
Date	Date	"date"	require("fashion-model/Date")
Boolean	Boolean	"boolean"	require("fashion-model/Boolean")
Number	Number	"number"	require("fashion-model/Number")
Integer		"integer"	require("fashion-model/Integer")
String	String	"string"	require("fashion-model/String")
Array	Array	"array" / []	require("fashion-model/Array")
Function	Function	"function"	require("fashion-model/Function")

Complex Object Type

Declare custom complex object type:

const Address = Model.extend({
  properties: {
    city: String,
    state: String,
    primary: Boolean,

    // The `{}` is equivalent to `"any"` data type
    metadata: {}
  }
});

Create instance via new constructor with no initial data:

// Create via constructor with no initial data
const address = new Address();
address.setCity('San Francisco');
address.setState('CA');
address.setPrimary(true);
address.setMetadata({a: 'b'});

Create instance via new constructor with some initial data:

// Create via constructor with initial data
const address = new Address({
  city: 'San Francisco',
  state: 'CA',
  primary: true,
  metadata: {
    a: 'b'
  }
});

Create instance via create method:

// Create via "create" function
const address = Address.create();
address.setCity('San Francisco');
address.setState('CA');
address.setPrimary(true);
address.setMetadata({a: 'b'});

Create instance by wrapping existing data:

// Create via "wrap" function.
// If the argument to `Address.wrap(...)` function is already
// the correct type then the original value will be returned.
address = Address.wrap({
  city: 'San Francisco',
  state: 'CA',
  primary: true,
  metadata: {
    a: 'b'
  }
});

Types that Implement EventEmitter

Types that extend Model will not implement the EventEmitter interface. If your type should be an EventEmitter then your type should either extend require('fashion-model/ObservableModel') or add the EventEmitter mixin. Types that implement EventEmitter will emit change and change:someProperty events

Example using ObservableModel:

const Something = require('fashion-model/ObservableModel').extend({
  properties: {
    value: String
  }
});

Example using mixin:

const Something = require('fashion-model/Model').extend({
  properties: {
    value: String
  },
  mixins: [require('fashion-model/mixins/EventEmitter')]
});

Listening for property value changes:

const something = new Something();

something.on('change:value', function (event) {
  // The "value" property changed
  console.log(
    'Old value: ' + event.oldValue,
    'New value: ' + event.newValue);
});

something.on('change', function (event) {
  // Some property changed
  console.log(
    'Property: ' + event.propertyName,
    'Old value: ' + event.oldValue,
    'New value: ' + event.newValue);
});

Self-type References in Properties

In some use cases, the type of a property is the same type as the complex object for which the property is declared. For exampled, to build a linked list, each node has a pointer to the next node.

Here are some examples of self-type references:

// Declare a linked list node type that has a pointer
// to the next node
const LinkedListNode = Model.extend({
  properties: {
    next: 'self',
    value: Object
  }
});

// Here is another functionally equivalent variation of LinkedListNode
const LinkedListNode = Model.extend({
  properties: {
    next: {
      type: 'self'
    },
    value: Object
  }
});

// An example of self-type reference within an array
const TreeNode = Model.extend({
  properties: {
    children: ['self'],
    value: Object
  }
});

// Here is another functionally equivalent variation of TreeNode
const TreeNode = Model.extend({
  properties: {
    // Brackets at end of type name are used to denote arrays
    children: 'self[]',
    value: Object
  }
});

Getters and Setters

A getter and setter will be generated on the prototype, for each property defined in the model.

For example:

// Define an Address model
const Address = Model.extend({
  properties: {
    city: String,
    state: String
  }
});

// Create instance of Address
const address = new Address();

// Use the generated setter to set the city
address.setCity('New York');

// Use the generated getter to get the city
assert(address.getCity() === 'New York')

Note: The getter function name will be always in the form get<PropertyName>. The setter function name will be always in the form set<PropertyName>. These rules do not change for properties with Boolean type.

Model prototype

The Model types are created via standard prototypical inheritance. If you wish to conveniently add other methods or properties to the prototype then use use the prototype property in the Model configuration.

For example:

const Person = Entity.extend({
  properties: {
    firstName: String,
    lastName: String
  },
  prototype: {
    getDisplayName: function () {
      return this.getFirstName() + ' ' + this.getLastName();
    }
  }
});

Inheritance

Define your base Entity type:

const Entity = Model.extend({
  properties: {
    id: String
  }
});

Define a type that extends Entity:

const Person = Entity.extend({
  properties: {
    email: String
  }
});

The new Person type will recognize email (defined for Person) and id (defined for Entity) as properties.

const person = new Person();
person.setId('john-doe');
person.setEmail('[email protected]');

You can also create getters for computed/derived properties.

For example:

const Person = Entity.extend({
  properties: {
    firstName: String,
    lastName: String,
    displayName: {
      type: String
      get: function (property) {
        return this.getFirstName() + ' ' + this.getLastName();
      }
    }
  }
});

Non-persisted Properties

If you'd like to store computed properties in the Model instance for performance reasons but you don't want them to be persisted to storage, then you might want to mark an property as non-persisted.

For example, here's a Model type that will automatically update displayName whenever firstName or lastName is changed:

function _updateDisplayName (person) {
    person.setDisplayName(person.getFirstName() + ' ' + person.getLastName());
}

const Person = Entity.extend({
  properties: {
    firstName: {
      type: String,
      set: function (value, property) {
        this.data[property.getKey()] = value;
        _updateDisplayName(this);
      }
    },
    lastName: {
      type: String,
      set: function (value, property) {
        this.data[property.getKey()] = value;
        _updateDisplayName(this);
      }
    },
    // displayName is updated whenever firstName or lastName change
    displayName: {
      type: String,

      // do not persist displayName when clean() is called since it is
      // a derived value
      persist: false
    }
  }
});

const person = new Person({
    firstName: 'John',
    lastName: 'Doe'
});

assert(person.getDisplayName() === 'John Doe');

// Remove non-persisted properties
const personObj = person.clean();
assert(personObj.displayName === undefined);

Wrap/Unwrap

Model.unwrap(obj) can used to return the underlying data for a Model instance. If the given obj is not a Model instance then obj is returned.

SomeType.wrap(obj) can be used to ensure that the given obj is wrapped as SomeType. If obj is already SomeType then obj will simply be returned.

Examples:

const Address = Model.extend({
  properties: {
    city: String,
    state: String
  }
});

const address = new Address({
  city: 'San Francisco',
  state: 'CA'
});

// Create an instance of Address
const addressObj = Model.unwrap(address);
assert(addressObj.city === 'San Francisco');

// Wrap the unwrapped object
const addressWrapped = Address.wrap(addressObj);
assert(addressWrapped.getCity() === 'San Francisco');

// The wrapped object returned by Address.wrap()
// will be the original Model instance that we created.
assert(addressWrapped === address);

Clean

Model.clean(obj) should be used to return a clone of an object in which all non-persisted properties and metadata have been removed. The clean function will always return a deep clone of the given object if the given argument is non-null and not a primitive.

const address = new Address({
  city: 'San Francisco',
  state: 'CA'
});

// When saving a model object to disk or storage, use clean to remove
// unnecessary fields.
db.save(address.clean(), callback);

A Model can also control how its data is cleaned by providing a clean property. For example, this might be helpful for working with binary data by automatically encoding the binary data as a base64 string.

A Model type that is not wrapped (that is, when wrap: false flag is provided), its value will not be cleaned unless a function is provided for the clean property.

Here's an example how to use the clean function to convert a Buffer to a Base64 encoded string:

const Binary = Model.extend({
  // Don't wrap binary data because we want to use the raw Buffer type
  // provided by Node.js runtime environment
  wrap: false,

  // Provide a clean function that will be used to clean values
  // associated with properties whose type is Binary
  clean: function (value) {
    // clean will convert to base64
    return value.toString('base64');
  },

  coerce: function (value, options) {
    if (value == null) {
        return value;
    }

    if (value.constructor === Buffer) {
        // no conversion needed
        return value;
    }

    // Buffers can be of type array. We assume that if an array is provided,
    // that it is an array of bytes.
    if (Array.isArray(value)) {
        return new Buffer(value);
    }

    if (value.constructor === String) {
        // assume that a string represents base64 encoded data
        return new Buffer(value, 'base64');
    }

    this.coercionError(value, options, 'Invalid binary data.');
  }
});

const Image = Model.extend({
  properties: {
    data: Binary
  }
});

const image = new Image({
  // data can be provided as Array of bytes, base64 encoded string, or Buffer
  // because Binary.coerce function handles each of these.
  data: someData
});

// the data will be converted to Buffer object via Binary.coerce function
assert(image.getData() instanceof Buffer);

// Calling clean on the image model instance will cause the contained data
// to be converted to base64 string via Binary.clean function.
const cleanedImage = image.clean();

// the data will be converted to String via Binary.clean function
assert(typeof cleanedImage.data.constructor === 'string');

// log the Base64 encoded string
console.log(cleanedImage.data);

Implementing a clean function on individual properties is also supported:

const Person = Model.extend({
  properties: {
    name: String,
    ssn: {
      type: String,
      clean: function (value, options) {
        // Only return the value of the ssn property if the `showSensitive`
        // option is passed to `clean`.
        return (options.showSensitive) ? value : undefined;
      }
    }
  }
});

const person = new Person({ name: 'John', ssn: 'abc123' });

assert.deepEqual(person.clean(), { name: 'John' });
assert.deepEqual(person.clean({ showSensitive: true }), {
  name: 'John',
  ssn: 'abc123'
});

Stringify

Model instances have a stringify function that can be used to safely stringify the instance.

For example:

// Stringify and do not add extra white-space
console.log(model.stringify());

// Stringify and include extra white-space for better readability
console.log(model.stringify(true));

Type Coercion

As a developer, you may choose to be lenient about how certain non-Model instances are coerced into instances of a Model.

For example, consider this example of declaring ObjectId type that automatically coerces Strings to actual instances of require('mongodb').ObjectID:

const MongoDbObjectID = require('mongodb').ObjectID;

const ObjectId = Model.extend({
  // Don't wrap object ID.
  // This means that the getters for properties of this type
  // will return the raw MongoDB ObjectID type
  wrap: false,

  // We provide a "coerce" function to convert a value to the proper
  // MongoDB ObjectID type
  coerce: function (data) {
    if (data == null) {
      return data;
    } else {
      // Use the MongoDB ObjectID constructor to coerce our
      // value (for example, it will handle String instances)
      return new MongoDbObjectID(data);
    }
  }
});

const Entity = Model.extend({
  id: {
    // ObjectId is a type that we use just to make sure that the value
    // is automatically converted to the type that we need for storage
    type: ObjectId,

    // MongoDB data storage expects a document to store its
    // identifier in the "_id" property but we still want to
    // access it via "getId" and "setId" (and not "get_id" and "set_id")
    key: '_id'
  }
})

Models that use the primitive Date type also benefit from type coercion. The Date coerce function provided by fashion-model automatically convert strings in ISO date format to Date instances. You will probably find this helpful because, by default, JSON.stringify(obj) will automatically convert Date objects to Strings using the standard ISO format.

For example:

const Document = Model.extend({
  dateCreated: Date
});

const document = new Document();
// A String value in ISO date format is automatically converted to a
// real Date.
document.setDateCreated('2014-12-22T21:18:45.905Z');

Enum Type

Short-hand syntax for declaring an enum type:

const Color = Enum.create(['red', 'green', 'blue']);

Alternate syntax for declaring an enum type:

const Color = Enum.create({
  values: ['red', 'green', 'blue']
});

Enum type access patterns:

Based on the Color enum type declared in examples above, the constant enum values will be accessible from the new Color type. Each Color enum value will have some helper functions as shown shown in the examples below.

// The following assertions will be true
// test to see

// Check to see if given color is the RED enum value
const color = Color.RED;
assert(color.isRed());

// Get the name of the enum
assert(Color.RED.name() === 'red');

// Get the value associated with the enum
assert(Color.RED.value() === 'red');

// Clean the enum value which will return its name
assert(Color.RED.clean() === 'red');

// Get the ordinal value associated with index of the value.
assert(Color.RED.ordinal() === 0);

Object enum values:

const Color = Enum.create({
  values: {
    red: {
      hex: '#FF0000',
      name: 'Red'
    },

    green: {
      hex: '#00FF00',
      name: 'Green'
    },

    blue: {
      hex: '#0000FF',
      name: 'Blue'
    }
  }
});

// The following assertions will be true:
assert(Color.red.name() === 'red');
assert(Color.red.value().hex === '#FF0000');
assert(Color.red.value().name === 'Red');
assert(Color.RED.name() === 'red');
assert(Color.RED.value().hex === '#FF0000');
assert(Color.RED.value().name === 'Red');
assert(Color.RED.ordinal() === 0);

Loop over values:

Color.values.forEach(function (colorValue) {
  console.log('Color ' + colorValue.name());
});

Loop over names:

Color.names.forEach(function (colorName) {
  console.log('Color ' + colorName);
});

Array Type

Syntax:

const Color = Enum.create({
  values: ['red', 'green', 'blue']
});

const ColorPalette = Model.extend({
  properties: {
    colors: {
      // colors has type array
      type: Array,

      // each item in the array is a Color
      items: Color
    }
  }
});

Short-hand syntax:

const ColorPalette = Model.extend({
  properties: {
    // Using an Array instance is short-hand for specifying
    // that the property is of type array. The first item
    // in this array indicates the type of each item.
    colors: [Color]
  }
});

Accessing an array property:

const colorPalette = new ColorPalette({
  colors: ['red', 'green', 'blue']
});

// getColors() will return an Array and we can use the "forEach" function.
// Each item in the returned Array will be an instance of Color.
colorPalette.getColors().forEach(function (color, index) {
  assert(color.constructor === Color);
});

Object Validation

Using array to capture errors:

// array that will collect errors
const errors = [];

// collect errors while wrapping existing person data
const person1 = Person.wrap({
  name: 'John',
  age: 'bad integer'
}, errors);

// collect errors while constructing new person
const person2 = new Person({
  name: 'John',
  age: 'bad integer'
}, errors);

Using extended options:

const options = {
  // array that will collect errors
  errors: [],

  // strict mode is used by some primitive types to require
  // that values be of the same primitive types
  // (no automatic type coercion)
  strict: true
};

const person = new Person({
  name: 'John',
  age: 'bad integer'
}, options);

JSON Schema

A Model type can be easily converted to an equivalent JSON schema with the following module:

const jsonSchema = require('fashion-model/json-schema-draft4');
const someModelSchema = jsonSchema.fromModel(SomeModel, options);

Option	Type	Purpose
toRef	function (Model)	This function can be used to turn a Model definition to a reference name (return value will be used as value for $ref properties)
isIgnoredProperty	function (name, property)	This function can be used to exclude a property from the schema definition of a complex object

Convert Model to JSON Schema

Define your models:

const Model = require('fashion-model/Model');
const Enum = require('fashion-model/Enum');

const Entity = Model.extend({
  typeName: 'Entity',
  properties: {
    id: String
  }
});

const Gender = Enum.create({
  typeName: 'Gender',
  title: 'Gender',
  description: 'A person\'s gender',
  values: ['M', 'F']
});

const Species = Enum.create({
  typeName: 'Species',
  title: 'Species',
  description: 'A species',
  values: ['dog', 'cat']
});

const Pet = Model.extend({
  typeName: 'Pet',
  properties: {
    name: String,
    species: Species
  }
});

const Person = Entity.extend({
  typeName: 'Person',
  title: 'Person',
  description: 'A person',
  properties: {
    name: String,
    dateOfBirth: Date,
    gender: Gender,
    age: 'integer',
    pets: [Pet],
    favoriteNumbers: ['integer'],
    anything: [],
    blob: Object
  }
});

Convert your model to JSON schema:

const jsonSchema = require('fashion-model/json-schema-draft4');
const jsonSchemaOptions = {
  toRef: function (Model) {
    return Model.typeName;
  }
};

const EntitySchema = jsonSchema.fromModel(Entity, jsonSchemaOptions);
const GenderSchema = jsonSchema.fromModel(Gender, jsonSchemaOptions);
const SpeciesSchema = jsonSchema.fromModel(Species, jsonSchemaOptions);
const PetSchema = jsonSchema.fromModel(Pet, jsonSchemaOptions);
const PersonSchema = jsonSchema.fromModel(Person, jsonSchemaOptions);

Entity JSON Schema:

{
  "id": "Entity",
  "type": "object",
  "properties": {
    "id": {
      "type": "string"
    }
  }
}

Gender JSON Schema:

{
  "id": "Gender",
  "title": "Gender",
  "description": "A person's gender",
  "type": "string",
  "enum": [
    "M",
    "F"
  ]
}

Species JSON Schema:

{
  "id": "Species",
  "title": "Species",
  "description": "A species",
  "type": "string",
  "enum": [
    "dog",
    "cat"
  ]
}

Pet JSON Schema:

{
  "id": "Pet",
  "type": "object",
  "properties": {
    "name": {
      "type": "string"
    },
    "species": {
      "$ref": "Species"
    }
  }
}

Person JSON Schema:

{
  "id": "Person",
  "title": "Person",
  "description": "A person",
  "allOf": [
    {
      "$ref": "Entity"
    }
  ],
  "type": "object",
  "properties": {
    "name": {
      "type": "string"
    },
    "dateOfBirth": {
      "type": "string",
      "format": "date-time"
    },
    "gender": {
      "$ref": "Gender"
    },
    "age": {
      "type": "integer"
    },
    "pets": {
      "type": "array",
      "items": {
        "$ref": "Pet"
      }
    },
    "favoriteNumbers": {
      "type": "array",
      "items": {
        "type": "integer"
      }
    },
    "anything": {
      "type": "array"
    },
    "blob": {
      "type": "object"
    }
  }
}