Define Model
and Collection
like types to manage your Immutable
data structures in a functional way
Immutable.js gives us great basic data constructs like Map
, List
and Set
. Vry makes it straightforward to define types that house common logic for managing your immutable data. Types, generated of templates like Model
, Collection
and Ref
are objects of stateless functions. This means that the instances (Immutable.Maps) are passed to the Type's methods as the first argument and a new / updated version is returned. This makes them a great fit to implement Redux reducers.
The project is not yet complete, with constructs being added as I find a good way to express and abstract them. While still somewhat experimental the basic behaviours have been developed throughout a sequence of various projects, so semver will be respected.
Documentation so far is limited to an API reference, which should be enough to get started, but I'd be happy to add guides.
Use in your project:
npm install --save vry
To run tests, in vry root run:
npm install --dev
npm test
const Invariant = require('invariant')
const Immutable = require('immutable');
const { State } = require('vry')
// define entities, by name and with defaults
const User = State.create('user', {
id: null
email: null,
name: ''
})
// the factory accepts attributes and returns an instance
const homer = User.factory({
name: 'Homer Simpson'
})
Invariant(Immutable.Map.isMap(homer), 'instance is an Immutable.Map, plain and simple')
// add your own methods
User.hasEmail = function(user) {
// make sure an actual user was passed
Invariant(User.instanceOf(user), 'User required to check whether user has an email')
return !!user.get('email')
}
const Post = State.create('post', {
title: 'Untitled',
author: null
})
const homersPost = Post.factory({
// nest entities
author: homer,
// use any type of Immutable.Iterable
tags: Immutable.Set(['homer', 'springfield', 'yellow'])
})
// serialize
const rawPost = Post.serialize(homersPost);
// anaemic models are great when combined with functional programming paradigms
const users = Immutable.List([homer]);
const usersWithEmails = users.filter(User.hasEmail);
Working with a single state tree for your application state (like Redux) is great, but requires changes to be immutable. Using Immutable.js makes for a great fit, however, all you get are basic constructs like Map
, List
, Set
. And while there is a Record
construct, it still means that a lot of generic behaviour for entities has to be written by hand. Through a sequence of projects this represents the basics that keep coming back.
Among plenty of other ideas, these are constructs I've validated the need for within my own projects:
Collection
abstracting basic collection behaviour like storing lists of entities under multiple indexes, merging behaviour, etc.NormalisedGraph
with a standard way of storing and referring to instances of models
State
- the most basic type generator in VryModel
- likeState
but with better support for nesting using aSchema
Schema
- to specify how various types compose togetherRef
- aState
type to make refering to other state easier
State is the most basic type in Vry. It allows you to create instances of a type with defaults in place, identify instances, merge them and serialize them back to plain javascript objects.
Create a new type of State. Returns an object with methods to work with the created state type.
name
- (required) name of the state typedefaults
- plain object orImmutable.Iterable
of default key-value pairs that are used as the base of every instance
const Vry = require('vry')
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
Note: All methods are bound to the state itself, so you can call them without binding them manually. For the unbound versions, see the prototype of the created type (state.prototype
). Methods added after creation are not bound automatically.
Returns a boolean indicating whether the passed value is an instance of any state type. It's useful where Immutable.Map.isMap
is not sufficient for your logic, for example when you need to know the meta attributes are present.
maybeState
- any value
Note: with the current implementation a Model
instance will also pass this test.
const notStateUser = Immutable.Map({ name: 'Homer '})
const user = User.factory(notStateUser)
console.log(State.isState(notStateUser)) // false
console.log(State.isState(user)) // true
Returns the defaults as defined with State.create
, represented as an Immutable.Map
.
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
const defaults = User.defaults()
console.log(defaults)
// Immutable.Map {
// name: 'New user',
// email: null,
// activated: false
// }
Returns the name (string) as defined with State.create
.
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
const name = User.typeName()
console.log(name) // 'user'
Returns a new instance (Immutable.Map
) of the state type using the type's defaults as a base and populated with the passed attributes. It also adds some meta data to keep track of the instance identity.
attributes
-object
or anyImmutable.Iterable
of key-value pairs with which the type defaults will be overridden and amended. Nestedobject
s andarray
s are converted toImmutable.Map
andImmutable.List
respectively, while anyImmutable.Iterable
s will be left untouched.options
-object
with the following keysparse
-function
as described bystate.parse
that is to be used instead ofstate.parse
to transform the passed inattributes
.defaults
- plainobject
orImmutable.Iterable
of default key-value pairs that are used as the base of the instance, instead of those defined for theState
.
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
const user = User.factory({
name: 'Homer'
})
console.log(user)
// Immutable.Map {
// name: 'Homer',
// email: null,
// activated: false
// }
Returns a boolean indicating whether the passed value is an "instance" of this state.
maybeInstance
- any value
const user = User.factory()
const post = Post.factory()
console.log(User.instanceOf(user)) // true
console.log(User.instanceOf(post)) // false
Returns a boolean indicating whether the passed value is an Immutable.Collection
(and so by extension things like Immutable.List
, Immutable.Set
, etc) of which all values are an instance of that state. Basically collection.every(state.instanceOf)
.
maybeCollection
- any value
const users = Immutable.List([
User.factory(),
User.factory()
])
const posts = Immutable.List([
Post.factory(),
Post.factory()
])
const mixed = Immutable.List([
User.factory(),
Post.factory()
])
console.log(User.collectionOf(users)) // true
console.log(User.collectionOf(posts)) // false
console.log(User.collectionOf(mixed)) // false
A place to implement custom parsing behaviour. This method gets called by state.factory
unless it got called with a parse
option, in which case state.parse
will be ignored.
Must return a Immutable.Iterable
. Any Immutable.Seq
s returned are converted into Immutable.Map
and Immutable.List
. By default this method is a no-op, simply returning the attributes passed in.
attributes
- (required)Immutable.Map
of attributes. Any plainobject
s orarray
s are represented asImmutable.Seq
s (Keyed
andIndexed
respectively), making it easy to deal with nested collections with a uniform API and giving you the opportunity to convert them to something else like aSet
.options
-object
with the following keysdefaults
- plainobject
orImmutable.Iterable
of default key-value pairs that are used as the base of the instance, instead of those defined for theState
.
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
User.parse = (attributes, options) => {
// Make sure names start with a capital
const name = attributes.get('name')
return attributes.set('name', name.charAt(0).toUpperCase() + name.slice(1))
}
const user = User.factory({
name: 'homer'
})
console.log(user.get('name')) // "Homer"
Returns the passed instance as a plain object.
instance
- (required)Immutable.Iterable
that represents an instance created withstate.factory
options
-object
with the following keysomitMeta
- when falsey the identity meta data will be included in the result. Defaults totrue
const user = User.factory({
name: 'Homer'
})
const plainUser = User.serialize(user)
console.log(plainUser)
// {
// name: 'Homer',
// email: null,
// activated: false
// }
Merges anything from plain attributes to other instances (even of a different type) with an existing instance. Any attributes present in the source
will override the ones in target
, except for the meta data of target
.
target
- (required)Immutable.Iterable
that represents an instance created withstate.factory
which will be the target of this mergesource
- (required)object
orImmutable.Iterable
of attributes to be merged with the target
const homer = User.factory({
name: 'Homer'
})
const activatedHomer = User.merge(homer, {
activated: true
})
console.log(activatedHomer)
// Immutable.Map {
// name: 'Homer',
// email: null,
// activated: true
// }
The Model
is a lot like State
, but goes beyond the basics of just maintaining an entity's identity and defaults. So far that means enhanced parsing and serialising of nested models within other models by use of a Schema
. Soon merging will join that club too, as well the concept of comparing instances by id attributes. Any other enhancements to working with entities that go beyond the logic of a single depth entity (State
) that we'll make in the future will probably end up on Model
too.
Create a new type of Model. Returns an object with methods to work with the created model type.
spec
- (required) string or object - either a string with the name of the model type or an object with values for the following properties:typeName
- (required) name of the model typedefaults
- plain object orImmutable.Iterable
of default key-value pairs that are used as the base of every instance. Same asdefaults
argument forState.create(name, defaults)
schema
- schema definition that describes any nested models. See the documentation forSchema
.
const Vry = require('vry')
const User = Vry.Model.create({
typeName: 'user',
defaults: {
name: 'New user',
email: null,
activated: false
}
})
const Post = Vry.Model.create({
typeName: 'post',
defaults: {
title: 'Untitled post',
body: '',
author: null
},
schema: {
author: User
}
})
Note: All methods are bound to the model itself, so you can call them without binding them manually. For the unbound versions, see the prototype of the created type (model.prototype
). Methods added after creation are not bound automatically.
Returns a boolean indicating whether the passed value is an instance of any model type. It's useful where Immutable.Map.isMap
is not sufficient for your logic, for example when you need to know the meta attributes are present.
Note: with the current implementation a State
instance will also pass this test.
maybeModel
- any value
const notStateUser = Immutable.Map({ name: 'Homer '})
const user = User.factory(notStateUser)
console.log(Model.isModel(notStateUser)) // false
console.log(Model.isModel(user)) // true
Returns the defaults as defined with Model.create
, represented as an Immutable.Map
.
const User = Vry.Model.create({
typeName: 'user',
defaults: {
name: 'New user',
email: null,
activated: false
}
})
const defaults = User.defaults()
console.log(defaults)
// Immutable.Map {
// name: 'New user',
// email: null,
// activated: false
// }
Returns the name (string) as defined with Model.create
.
const User = Vry.Model.create({
typeName: 'user',
defaults: {
name: 'New user',
email: null,
activated: false
}
})
const typeName = User.typeName()
console.log(typeName) // 'user'
Returns a new instance (Immutable.Map
) of the model using the model's defaults as a base and populated with the passed attributes. It also adds some meta data to keep track of the instance identity and uses the defined schema to defer creating nested models with their own factory
methods.
attributes
-object
or anyImmutable.Iterable
of key-value pairs with which the type defaults will be overridden and amended. The model's schema is used to handle the creation of nested types. Nestedobject
s andarray
s are converted toImmutable.Map
andImmutable.List
respectively, while anyImmutable.Iterable
s will be left untouched.options
-object
with the following keysparse
-function
as described bystate.parse
that is to be used instead ofmodel.parse
to transform the passed inattributes
.defaults
- plainobject
orImmutable.Iterable
of default key-value pairs that are used as the base of the instance, instead of those defined for the model. Nested values are propagated to nested models.schema
- schema definition that describes any nested models, to be used instead of the schema defined for the model. See the documentation forSchema
.
const User = Vry.Model.create({
typeName: 'user',
defaults: {
name: 'New user',
email: null,
activated: false
}
})
const Post = Vry.Model.create({
typeName: 'post',
defaults: {
title: 'Untitled post',
body: '',
author: null
},
schema: {
author: User
}
})
const post = Post.factory({
title: 'A Totally Great Post',
author: {
name: 'Homer'
}
})
console.log(post)
// Immutable.Map {
// title: 'A Totally Great Post',
// body: '',
// author: Immutable.Map {
// name: 'Homer',
// email: null,
// activated: false
// }
// }
Returns a boolean indicating whether the passed value is an "instance" of this type.
maybeInstance
- any value
const user = User.factory()
const post = Post.factory()
console.log(User.instanceOf(user)) // true
console.log(User.instanceOf(post)) // false
Returns a boolean indicating whether the passed value is an Immutable.Collection
(and so by extension things like Immutable.List
, Immutable.Set
, etc) of which all values are an instance of that type. Basically collection.every(model.instanceOf)
.
maybeCollection
- any value
const users = Immutable.List([
User.factory(),
User.factory()
])
const posts = Immutable.List([
Post.factory(),
Post.factory()
])
const mixed = Immutable.List([
User.factory(),
Post.factory()
])
console.log(User.collectionOf(users)) // true
console.log(User.collectionOf(posts)) // false
console.log(User.collectionOf(mixed)) // false
A place to implement custom parsing behaviour. This method gets called by type.factory
unless it got called with a parse
option, in which case type.parse
will be ignored.
Must return a Immutable.Iterable
. Any Immutable.Seq
s returned are converted into Immutable.Map
and Immutable.List
.
By default it uses the Schema
of the model to defer the parsing of other nested models to their own methods. If the passed value is a Ref
instance, the Schema
's factory
is ignored.
attributes
- (required)Immutable.Map
of attributes. Any plainobject
s orarray
s are represented asImmutable.Seq
s (Keyed
andIndexed
respectively), making it easy to deal with nested collections with a uniform API and giving you the opportunity to convert them to something else like aSet
.options
-object
with values for the following keysschema
- schema definition that describes any nested models, to be used instead of the schema defined for the model. See the documentation forSchema
.defaults
- plainobject
orImmutable.Iterable
of default key-value pairs that are used as the base of the instance, instead of those defined for theModel
. Nested values are propagated to nested models.
const User = Vry.Model.create({
typeName: 'user',
defaults: {
name: 'New user',
email: null,
activated: false
}
})
User.parse = (attributes) => {
// Make sure names start with a capital
const name = attributes.get('name')
return attributes.set('name', name.charAt(0).toUpperCase() + name.slice(1))
}
const Post = Vry.Model.create({
typeName: 'post',
defaults: {
title: 'Untitled post',
body: '',
author: null
},
schema: {
author: User
}
})
const post = Post.factory({
title: 'Nice post',
author: {
name: 'homer'
}
})
console.log(post.get('user').get('name')) // "Homer"
Returns the passed instance as a plain object. Defers the serialising of nested types to their serialize
methods when defined (see Schema
). Any references (instances of Ref
) are serialized as one.
instance
- (required)Immutable.Iterable
that represents an instance created withstate.factory
options
-object
with the following keysomitMeta
- when falsey the identity meta data will be included in the result. Defaults totrue
schema
- schema definition that describes any nested models, to be used instead of the schema defined for the model. See the documentation forSchema
.
const user = User.factory({
name: 'Homer'
})
const plainUser = User.serialize(user)
console.log(plainUser)
// {
// name: 'Homer',
// email: null,
// activated: false
// }
Merges anything from plain attributes to other instances (even of a different type) with an existing instance. Any attributes present in the source
will override the ones in target
, except for the meta data of target
.
target
- (required)Immutable.Iterable
that represents an instance created withmodel.factory
which will be the target of this mergesource
- (required)object
orImmutable.Iterable
of attributes to be merged with the target
const homer = User.factory({
name: 'Homer'
})
const activatedHomer = User.merge(homer, {
activated: true
})
console.log(activatedHomer)
// Immutable.Map {
// name: 'Homer',
// email: null,
// activated: true
// }
A Schema
is an object to describe how other types are embedded within a given type. It specifies the shape, as well as how the embedded types should be created, identified and serialized back to plain objects and arrays.
Each Model
type has one, detailing what other Model
types are embedded within it.
Schema's are plain javascript objects with either a type definition (see Schema.isType
) or a nested schema specified as values. There isn't a Schema.create
method to create schemas (yet).
// user has no embedded types
const postSchema = {
title: {
factory(value) { return value.toUpperCase() },
serialize(value) { return value.toLowerCase() }
},
author: User, // Types created with `Model.create` and `State.create` are valid type definitions!
// use `Schema.listOf` and `Schema.setOf` to convienently create types for nested
comments: Schema.listOf(Comment),
metadata: { // schemas can be nested
tags: Schema.setOf(Tag)
}
}
Returns whether a value is a valid schema definition. A valid schema is an object
with either a valid type definition (see Schema.isType
) or a nested schema.
maybeSchema
- any value
const postSchema = {
title: { // valid type definition
factory(value) { return value.toUpperCase() },
serialize(value) { return value.toLowerCase() }
}
}
const userSchema = {
name: "woohoo" // not a valid type definition
}
console.log(Schema.isSchema(postSchema)) // true
console.log(Schema.isSchema(userSchema)) // false
Returns whether a value is a valid type definition.
maybeType
- any value
A valid type definition is an object
with at least one of the following keys defined as described:
factory
- function with the signaturevar instance = function(plainValue, options)
serialize
- function with the signaturevar plainValue = function(isntance, options)
The following keys can be defined as well
instanceOf
- function with signaturevar isInstance = function(maybeInstance)
Returns a schema definition (IterableSchema
) describing an Immutable.List
of a schema or type.
definition
- either a valid schema or type definition
Returns a schema definition (IterableSchema
) describing an Immutable.Set
of a schema or type.
definition
- either a valid schema or type definition
Returns a schema definition (IterableSchema
) describing an Immutable.OrderedSet
of a schema or type.
definition
- either a valid schema or type definition
Returns whether the value is a special IterableSchema
, which are the special schema definitions returned by methods like Schema.listOf
.
maybeSchema
- any value
A Ref
, short for reference, is a bit of state that points to another bit of state. In practice it's used to allow for a single instance of an entity to be stored in one place, with any other uses of it referring back to the original. It's implemented as a State
type using State.create
, meaning that a reference is just another Immutable.Map
of a specific shape.
A common example of a ref in a data model is something like a user_id
inside a Post
to point the author by the id string of a User
. However, when reading the post, the user will have to be looked up separately. Using a Ref
instead of a plain id allows for automatic resolving of the user.
As we've implemented it so far the Ref
points to something by use of a path. What this path is relative to is completely flexible and up to you. That makes it a pretty low-level construct. However, I do have the intention to implement Ref
s based on a type definition and id instead, as part of the efforts for a Graph
construct.
Returns a Immutable.Map
that represents the Ref
with the specified path and also qualifies as a State
instance.
path
- (required) either a string, or an array /Immutable.List
of strings, specifying the path to target
const Vry = require('vry')
const User = Vry.State.create('user', {
name: 'New user',
email: null,
activated: false
})
const Post = Vry.State.create('post', {
title: 'New post',
author: null
})
const appState = Immutable.Map({
usersById: Immutable.Map({
3: User.factory({ name: 'Homer' }),
5: User.factory({ name: 'Marge' })
})
})
const post = Post.factory({
title: 'Keeping skyscrapers from collapsing',
author: Ref.create(['usersById', 3]) // choosing `appState` as the root of this reference
})
Note: a instance of Ref
, just like a relative path, holds no reference to what it's relative to. It's up to the user to apply it in the right way.
Returns the value the ref
is pointing to inside the source
. Returns undefined
when the path inside source
does not exist.
ref
- (required) instance of theRef
to be resolvedsource
- (required)Immutable.Iterable
to look up theref
in
const appState = Immutable.Map({
usersById: Immutable.Map({
3: User.factory({ name: 'Homer' }),
5: User.factory({ name: 'Marge' }),
6: User.factory({ name: 'Bart' })
})
})
const post = Post.factory({
title: 'Keeping skyscrapers from collapsing',
author: Ref.create(['usersById', 3]) // choosing `appState` as the root of this reference
})
const postAuthor = Ref.resolve(post.get('author'), appState)
console.log(postAuthor)
// Immutable.Map {
// name: 'Homer',
// ...
// }
Resolves all the Ref
instances inside an Immutable.Collection
using a source
.
refs
- (required)Immutable.Collection
containing onlyRef
instancessource
- (required)Immutable.Iterable
to look up therefs
in
const appState = Immutable.Map({
usersById: Immutable.Map({
3: User.factory({ name: 'Homer' }),
5: User.factory({ name: 'Marge' }),
6: User.factory({ name: 'Bart' })
})
})
const post = Post.factory({
title: 'Keeping skyscrapers from collapsing',
likers: [
Ref.create(['usersById', 5]), // choosing `appState` as the root of this reference
Ref.create(['usersById', 6])
]
})
const postLikers = Ref.resolveCollection(post.get('likers'), appState)
console.log(postLikers)
// Immutable.List [
// Immutable.Map {
// name: 'Marge',
// ...
// },
// Immutable.Map {
// name: 'Bart',
// ...
// }
//
// ]
Returns an updated subject
in which all references encountered at the given paths
inside the subject
are resolved against a single source. When the path can not be found inside the subject
, it's ignored. When a reference
along a path
can not be resolved, the rest of that path is ignored.
source
- (required)Immutable.Iterable
the source in which to look up any encountered referencessubject
- (required)Immutable.Iterable
the subject for which to resolve any references...paths
- (required) string or array /Immutable.Iterable
of strings that represent a path insubject
. Paths are traversed up to the point that it exists. The rest is ignored
const appState = Immutable.Map({
usersById: Immutable.Map({
3: User.factory({
name: 'Homer',
friends: [
Ref.create(['usersById', 5]),
Ref.create(['usersById', 6])
]
}),
5: User.factory({ name: 'Marge' }),
6: User.factory({ name: 'Bart' })
})
})
const post = Post.factory({
title: 'Keeping skyscrapers from collapsing',
author: Ref.create(['usersById', 3]),
likers: [
Ref.create(['usersById', 5]), // choosing `appState` as the root of this reference
Ref.create(['usersById', 6])
]
})
const resolvedPost = Ref.replaceIn(appState, post, ['author', 'friends'], 'likers')
console.log(resolvedPost)
// Immutable.Map {
// likers: Immutable.List [
// Immutable.Map {
// name: 'Marge',
// ...
// },
// Immutable.Map {
// name: 'Bart',
// ...
// }
// ],
//
// author: Immutable.Map {
// name: 'Homer',
// friends: Immutable.List [
// Immutable.Map {
// name: 'Marge',
// ...
// },
// Immutable.Map {
// name: 'Bart',
// ...
// }
// ],
// ...
// }
// }
//