Redux consumer toolkit allows you to write your business logic and then embed it in a reducer or selector. The underlying business logic need not know anything about the reducer or selector it's embedded in. For example, here's one way to encapsulate how to get the currently signed in user.
const reducerState = {
authInfo: {
currentUserId: 1,
lastSignedInAt: '2017-09-17'
},
userData: {
1: { /* user id 1 data */ },
4: { /* user id 2 data */ },
10: { /* user id 10 data */ }
}
};
// Encapsulated from the knowledge of selectors!
const getUser = (userId, userData) => {
return userData[userId];
};
// Generic selector
const currentUserIdSelector = (state) => {
return state.authInfo.currentUserId;
};
// Generic selector
const userDataSelector = (state) => {
return state.userData;
};
// Just take the business logic and give it its data dependencies!
const currentUserSelector =
apAll(getUser, currentUserIdSelector, userDataSelector);The benefits of this decoupling come into play when you consider reusability.
Now I need to get a user id from component props. I can simply reuse the
getCurrentUser function with a different selector!
const userIdSelector = (state, props) => {
return props.userId;
};
const userFromPropsSelector =
apAll(getUser, userIdSelector, userDataSelector);Write reducers that have well isolated responsibilities and then combine them together seamlessly. Even isolate reducers that depend on the value of other reducers! Disentangle your monolithic reducers and make them easier to test and maintain. For example, here's how I might encapsulate entangled responsibilities of a reducer that handles form data.
const initialState = {
name: {
dirty: false,
value: null
},
email: {
dirty: false,
value: null
}
};
const originalReducer = (state, action) {
switch(action.type) {
case FIELD_CHANGED:
const newField = {
value: action.value,
dirty: true
};
return {
...state,
[action.fieldName]: newField
};
case FIELD_CLEARED:
const newField = {
value: null,
dirty: false
};
return {
...state,
[action.fieldName]: newField
};
}
}The reducer is conflating three logical responsibilities by handling field values, dirty status, and state setting at the same time. Those can easily be encapsulated and recombined with a few functions from the toolkit.
const dirtyHandler = (state, action) => {
switch(action.type) {
case FIELD_CHANGED:
return true;
case FIELD_CLEARED:
return false;
}
}
const valueHandler = (state, action) => {
switch(action.type) {
case FIELD_CHANGED:
return action.value;
case FIELD_CLEARED:
return null;
}
const newFieldReducer = combine({
dirty: dirtyHandler,
value: valueHandler
});
const formReducer = chain(
newFieldReducer,
(newField) => (state, action) => {
return {
...state,
[action.fieldName]: newField
};
}
);All of this decoupling and encapsulation makes it so you can focus on one thing at a time. Write your business logic without needing to focus on the details of reducers. Combine your selectors without needing to focus on the underlying business logic. Even write a reducer without needing to focus on tangential responsibilities because you've isolated them from each other.
Make your reducers, and selectors, more testable and easier to reason about and maintain with Redux Consumer Toolkit!
The inspiration for this library was fantasyland-redux, only instead of basing it off of the fantasyland specification it is based off of the static-land specification. This made it simple to build this as a library instead of pinning it to a specific redux version.
See the Github Pages site for more usage examples and documentation on the library. It is a literate program from which this library's source code is constructed.
The library is organized internally into modules corresponding to the structures defined by the static-land spec. However it exposes a more idiomatic Javascript API than what is specified in static-land. This overview will go over the idiomatic API.
The library assumes a traditional reducer signature of:
// Flow type
type Reducer<action, input, output> = (input, action) => output;
// Traditional signature
function reducer(state, action) {
return state;
}and a traditional selector signature of:
// Flow type
type Selector<props, input, output> = (input, props) => output;
// Traditional signature
function selector(state, props) {
return state;
}It unifies them into a single type called a consumer:
// Flow type
type Consumer<Static, Input Output> = (input, static) => output;
// Traditional signature
function consumer(input, static) {
const output = input;
return output;
}Flow types are exported.
Both memoized and non-memoized versions of these modules are exported. Memoized
versions can be accessed by importing from 'redux-consumer-toolkit/memoized'
instead of 'redux-consumer-toolkit'.
Modifying the returned state from a consumer.
- Pull only the properties that a specific component requires
- Derive computed properties from a reducer's state object
import { map } from 'redux-consumer-toolkit';
function grandTotal(receipt) {
return receipt.shipping + receipt.tax + receipt.subtotal;
}
map(grandTotal, receiptReducer);map : ((OutA => OutB), Consumer<Static, In, OutA>) => Consumer<Static, In, OutB>)Modifying both the input state and the output state of a consumer.
- Glue consumers lower in the hierarchy to reducers higher in the hierarchy
- Embed consumers in a context like an object key or reversible transformation
import { mapInOut } from 'redux-consumer-toolkit';
import Ramda from 'ramda';
function grandTotal(receipt) {
return receipt.shipping + receipt.tax + receipt.subtotal;
}
// Returns { lineItems, receipt }
const orderReducer;
// Ramda.prop gets the named property off an object
mapInOut(Ramda.prop('receipt'), grandTotal, orderReducer);mapInOut : ((InA => InB), (OutA => OutB), Consumer<Static, InB, OutA>) => Consumer<Static, InA, OutB>
Embed a consumer in an object.
objectify takes a key and a consumer and generates a consumer which can take
an object with that key and return an object with that key. You can reimplement
combineReducers this way.
import { objectify, identity, expandAll } from 'redux-consumer-toolkit';
const combineReducers = (reducerSpec) => (
reducerSpec.map((r, k) => objectify(k, r))
.reduce(expandAll, identity)
)objectify : (string, Consumer<Static, In, Out>) => Consumer<Static, { string: In }, { string: Out }>Like mapInOut, but only change the input of a consumer.
mapIn : ((InA => InB), Consumer<Static, InB, Out>) => Consume<Static, InA, Out>
Take transformation functions you would use with map, but have multiple
arguments, and apply them to multiple consumer. Each consumer gets the same
action/props and input state.
import { apAll } from 'redux-consumer-toolkit';
const grandTotal = (shipping, tax, subtotal) => {
return shipping + tax + subtotal;
};
apAll(of(grandTotal), shippingReducer, taxReducer, subtotalReducer);
apAll : (Consumer<Static, In, (mixed => Out)>, ...Array<Consumer<Static, In, mixed>>) => Consumer<Static, In, Out>Provides a way to take non-consumer things like functions and values and use
them with apAll. It essentially creates a consumer that always returns the
same value. See the use case for apAll for an example.
constant: (Out) => Consumer<Static, In, Out>Provides a way to take two (or more) consumers and run them with the same action/props, one after the other. The output of the first consumer is used as input to the second consumer.
import { concatAll } from 'redux-consumer-toolkit';
// Returns line items state
const lineItemsReducer;
// Takes line items and returns a total
const subtotalReducer;
concatAll(lineItemsReducer, subtotalReducer);concatAll : (Consumer<Static, In, OutA>, Consumer<Static, OutA, OutB>) => Consumer<Static, In, OutB>Provides a special consumer that can be used as an identity when combining
reducers using concat. This is handy if you have a list of reducers you want
to reduce over and need an initial value.
identity : Consumer<Static, In, In>Provides a way to combine consumers in such a way that the second consumer's behavior can depend on the output of the first. Both consumers receive the same input and static values. This is useful for consumers that depend on the values of other consumers in addition to their own internal state. This is a powerful but fairly low-level ability. Therefore some convenience functions are specified here as well for common combinations.
import { chain, mapIn } from 'redux-consumer-toolkit';
import Ramda from 'ramda';
// This consumer depends on the router state to know when to load more products
const productListReducer = (router) => (state, action) => { ... };
chain((routerState) => (
mapIn(Ramda.prop('productList'), productListReducer(router))
), mapIn(Ramda.prop('router'), routerReducer));chain : (Consumer<Static, In, OutA>, OutA => Consumer<Static, In, OutB>) => Consumer<Static, In, OutB>expandAll handles the use case where you might want consumers operating on the
same state to each return separate keys of a resulting object.
import { expandAll } from 'redux-consumer-toolkit';
// Manages user account information
const userReducer;
// Manages cart information, the cart may be anonymous or registered to a user.
const cartReducer;
// We might want to have all the cart and user information available in the same
// object
expandAll(userReducer, cartReducer);expandAll : (...Consumer<Static, In, *>) => Consumer<Static, In, Out>combine works just like the traditional combineReducers function, but it can
also operate on selectors. It also lacks the beginner-friendly runtime checks
that combineReducers provides. It accepts an object whose values are
consumers, and returns a consumer that will generate an object with the same
keys whose values are the state returned from the individual consumers.
combine : (Object<string, Consumer<Static, In, *>>) => Consumer<Static, In, Out>There are some utility functions to help debug consumers.
Log the output of the consumer before returning it.
logConsumer('message', consumer)Enters a breakpoint before returning the output of the consumer for easy examination.
debugConsumer(consumer)Requires webpack, babel, and eslint to build. You can use a Nix shell to enter a development environment with those tools already in it or just use your own.