A statically-typed immutable update helper library.
For Hydux.
- Statically-typed
- Support class state
- Using constructor to create new instance
- Class can define a
shallowClone
method to customize shallow clone.
yarn add hydux-mutator # or npm i hydux-mutator
import { setIn, updateIn } from 'hydux-mutator'
class User {
name = 'aa'
age = 0
}
let user = new User()
user = setIn(user, _ => _.name, 'bb')
user = updateIn(user, _ => _.age, a => a + 10)
Note: The accessor lambda function only support plain object/array. e.g.
setIn(user, _ => _.name, 'a')
setIn(user, _ => _.teacher.name, 'a')
setIn(user, _ => _['tags'][0], 'a')
setIn(user, _ => _.tags[0], 'a')
Which not support:
- Map/Set/Third-party collection library
- Function calls
But how can I use it in these scenarios? The answer is nesting:
let state = {
userMap: new Map<string, User>(),
userObjMap: {} as {[key: string]: User},
}
let key = 'a'
state = updateIn(state, _ => _.userMap, map => (
map = new Map(map),
map.set(key, setIn(map.get(key), _ => _.name, 'new name')),
map
))
Because hydux-mutator get the deep key path by parsing the lambda string, we cannot get the dynamic path in the accessor scope, so we have to pass it to the function as ctx, see:
let state = {
userObjMap: {} as {[key: string]: User},
}
let key = 'some key'
state = setIn(state, _ => _.userObjMap[key], 'new name', [key])
NOTE: The order of ctx should be the same as the occurrence order of dynamic keys.
We also provide some immutable collections like immutable-js, which has O(1) - O(logN) performance for update operations.
- ImmuList: Under the hood is @funkia/list, a modified implemenation of RRB tree with pretty good performance.
- ImmuMap: Based on an OCaml's immutable balanced tree implementation.
- ImmuSet: Based on ImmuMap.
Note: Polyfill for Symbol.iterator
is required!
All these collections can work seamlessly with setIn/updateIn/getIn/unsetIn
functions!
import { setIn } from 'hydux-mutator'
import ImmuList from 'hydux-mutator/lib/collections/list'
const book = {
title: 'book1'
}
const state = {
list: new ImmuList([book, book, book])
}
setIn(state, _ => _.list.get(0).title, 'new title')
We also support fb's immutable-js or others contains .get(key: string | number)
and .set(key: string | number, value: any)
methods.
Immer like api.
Note: This is based on es6 Proxy, you might need a proxy polyfill if you want to use in es5 environment.
import immer from 'hydux-mutator/lib/immer'
const state = {
subState: {
nestedSubState: 1
}
}
const nextState = immer(state, (draft, state) => {
draft.subState.nestedSubState = state.subState.nestedSubState + 1
})
// nextState => {
// subState: {
// nestedSubState: 2
// }
// }
The main difference is monolite is using es6's Proxy<T>
under the hood, which might not support well in many browsers.
hydux-mutator
is implement by parsing lambda function's source string(fn.toString()
), this have better browser support. And the parsing part can easily be cached, which means it can have better performance.
What's more, hydux-mutator
support class state, which I rarely see in other immuatble update helpers.
import * as mutator from 'hydux-mutator'
class User {
constructor(name = '', age = 0) { // constructor should have an overload of zero parameters.
this.name = name
this.age = age
}
}
let state = {
user: new User()
}
state = mutator.setIn(state, _ => _.user.name, 'New Name')
state.user instanceof User // true!
- flow has bug in checking
setIn
, see: #5569
Start Suit: Set 1 key
immutable x 932,328 ops/sec ±14.93% (46 runs sampled)
seamless-immutable x 67,021 ops/sec ±16.99% (44 runs sampled)
timm x 881,627 ops/sec ±17.53% (62 runs sampled)
monolite x 151,347 ops/sec ±13.51% (46 runs sampled)
mutator x 219,049 ops/sec ±5.03% (71 runs sampled)
Fastest is immutable,timm
Start Suit: Set 2 key
immutable x 732,474 ops/sec ±9.49% (57 runs sampled)
seamless-immutable x 28,802 ops/sec ±12.65% (53 runs sampled)
timm x 715,487 ops/sec ±9.31% (55 runs sampled)
monolite x 97,454 ops/sec ±12.41% (55 runs sampled)
mutator x 147,161 ops/sec ±16.37% (53 runs sampled)
Fastest is immutable,timm
Start Suit: Set 5 key
immutable x 374,647 ops/sec ±13.05% (55 runs sampled)
seamless-immutable x 19,725 ops/sec ±8.40% (62 runs sampled)
timm x 217,508 ops/sec ±9.00% (43 runs sampled)
monolite x 80,403 ops/sec ±6.32% (69 runs sampled)
mutator x 111,625 ops/sec ±4.92% (65 runs sampled)
Fastest is immutable
Start Suit: Set 10 key
immutable x 257,998 ops/sec ±4.59% (69 runs sampled)
seamless-immutable x 11,238 ops/sec ±10.45% (59 runs sampled)
timm x 219,370 ops/sec ±8.29% (59 runs sampled)
monolite x 32,778 ops/sec ±7.16% (48 runs sampled)
mutator x 54,496 ops/sec ±8.72% (60 runs sampled)
Fastest is immutable
MIT