Simple alternative to redux-saga or redux-observable.
Rather than using generators or Observables, most common operations are defined with a simple config object.
npm install --save redux-orchestrateimport { createStore, applyMiddleware } from 'redux'
import orchestrate from 'redux-orchestrate'
import reducer from './reducers'
const processManager = [
// process manager logic
]
// pass rules directly to middleware
const store = createStore(reducer, applyMiddleware(orchestrate(processManager)))In case of action(s) X -> dispatch action(s) Y
const processManager = [
{
case: [
SEND_MESSAGE_BUTTON_CLICKED,
MESSAGE_INPUT_ENTER_KEY_PRESSED
],
dispatch: ADD_MESSAGE
}
]In case of action(s) X -> dispatch action(s) Y
In case of action(s) Y -> dispatch action(s) Z
const processManager = [
{
case: [
SEND_MESSAGE_BUTTON_CLICKED,
MESSAGE_INPUT_ENTER_KEY_PRESSED
],
dispatch: ADD_MESSAGE
},
{
case: ADD_MESSAGE,
dispatch: [
ANOTHER_ACTION,
ONE_MORE
]
}
]In case of action(s) X -> wait for k miliseconds -> dispatch action(s) Y
const processManager = [
{
case: [
SEND_MESSAGE_BUTTON_CLICKED,
MESSAGE_INPUT_ENTER_KEY_PRESSED
],
delay: 500
dispatch: ADD_MESSAGE
}
]In case of action(s) X -> debounce for k miliseconds -> dispatch action(s) Y
const processManager = [
{
case: [
SEND_MESSAGE_BUTTON_CLICKED,
MESSAGE_INPUT_ENTER_KEY_PRESSED
],
debounce: 500
dispatch: ADD_MESSAGE
}
]In case of action(s) X -> perform logic using orignal action and state -> dispatch action(s) Y
const processManager = [
{
case: [
SEND_MESSAGE_BUTTON_CLICKED,
MESSAGE_INPUT_ENTER_KEY_PRESSED
],
dispatch: (action, state) => {
if (state.canAddMessage) {
return { ...action, type: ADD_MESSAGE }
}
}
}
]In case of action(s) X -> make an ajax request ->
-> in case of success -> dispatch Y
-> in case of failure -> dispatch Z
const processManager = [
{
case: ADD_MESSAGE,
get: {
url: 'https://server.com',
onSuccess: MESSAGE_SENT,
onFail: MESSAGE_SENDING_ERROR,
}
}
]post request using action.payload:
const processManager = [
{
case: ADD_MESSAGE,
post: action => ({
url: 'https://server.com/new',
data: {
content: action.payload
},
onSuccess: { type: MESSAGE_SENT, id: action.id },
onFail: { type: MESSAGE_SENDING_ERROR, id: action.id }
})
}
]making use od res and err response object from onSuccess and onFail:
const processManager = [
{
case: ADD_MESSAGE,
post: action => ({
url: 'https://server.com/new',
data: {
content: action.payload
},
onSuccess: res => ({
type: MESSAGE_SENT,
dataFromRes: res.data
id: action.id
}),
onFail: err => ({
type: MESSAGE_SENDING_ERROR,
errorMessage: err.message
id: action.id
})
})
}
]In case of action(s) X -> make an ajax request ->
in case of action(s) Y -> cancel ajax request
const processManager = [
{
case: ADD_MESSAGE,
post: {
url: `http://server.com`,
cancelWhen: [
STOP_SENDING
],
onSuccess: MESSAGE_SENT
}
}
]Now let's say we need to implement an autocomplete feature. In short, these are feature requirements:
- Any time the user changes an input field, make a network request
- If network request is not completed, but user had changed the input field again, cancel the previous request
- Don't spam "suggestion server". Make the request when user had stopped typing, by debouncing its events.
const processManager = [
{
case: SEARCH_INPUT_CHARACTER_ENTERED, // in case user has changed an input field
debounce: 500, // wait for user to stop typing (debouncing by 500ms)
get: action => ({
url: `http://s.co/${action.payload}`, // make a get request to a "suggestion server"
cancelWhen: [
SEARCH_INPUT_CHARACTER_ENTERED, // in case user starts typing again, cancel request
SEARCH_INPUT_BLURED // in case user is not using an input field, cancel request
],
onSuccess: res => ({
type: AUTOCOMPLETE_SUGGESTION, // if query was successful, dispatch an event
payload: res.data
})
})
}
]const processManager = [
{
case: ADD_MESSAGE,
post: (action, state) => ({
url: 'https://chat.app.com/new',
data: {
content: action.payload
},
onSuccess: () => {
if (state.canMarkAsSent) {
return { ...action, type: MESSAGE_SENT }
} else {
return { ...action, type: FOR_SOME_REASON_THIS_IS_DISPATHCED }
}
}
})
},
{
case: FOR_SOME_REASON_THIS_IS_DISPATHCED
post: (action, state) => ({
url: 'https://what.is.happening',
data: {
content: action.payload
},
onSuccess: MESSAGE_SENT,
onFail: MESSAGE_SENDING_ERROR
})
}
]Sometimes you may wish to add rules dynamically after middleware has been applied:
const processManager = [
// initial rules
]
const orchestrateMiddleware = orchestrate(processManager)
const store = createStore(reducer, applyMiddleware(orchestrateMiddleware))
orchestrateMiddleware.addRules([
// additional rules added dynamically
])This middleware is not an attempt to solve all your problems. If you need to handle more complex async operations which are better solved by some other tools (generators, observables), then you should use middlewares that supports them or define your own (it's not that hard).
Also, don't forget that you can combine multiple middlewares.
Note: additional operators could be supported in the future (but only if they don't significantly complicate the existing API).
Yes.
redux-orchestrate uses axios for making network requests.
All options passed in request (or aliases like post, get, etc.) is mapped with axios request config
Config object which defines the middleware logic is here reffered as "process manager".
This term is borrowed from CQRS/ES terminology where the same concept is also referred as "saga" - "a piece of code that coordinates and routes messages between bounded contexts and aggregates".
Term "orchestrate" is used to reffer to a single, central point for coordinating multiple entities and making them less coupled.
This is a broad term, usually used in service-oriented arhitectures and compared with its opossite concept - "choreography"
orchestrate(processManager, options)
The main array of objects defining action coordination.
const processManager = [
{
case: [
IN_CASE_THIS_EVENT_IS_DISPATCHED,
OR_THIS_EVENT
],
dispatch: DISPATCH_THAT_EVENT,
debounce: 500,
delay: 500,
request: {
method: 'get',
url: 'url',
cancelWhen: [
IF_REQUEST_IS_PENDING_CANCEL_IT_WHEN_THIS_IS_DISPATCHED,
OR_THIS
],
onSuccess: DISPATCH_THIS_IF_AJAX_SUCCEDED
onFail: DISPATCH_THIS_IF_AJAX_FAILED,
// other axios props
}
}
]Proceed with dispatching or making a request if action type is matched with the one defined in case.
{
// string
case: 'EVENT',
// array
case: [
'EVENT_1',
'EVENT_2'
],
// function
case: (action, state) => `PREFIX_${action.type}`
}Synchronously dispatch an action
{
// string
dispatch: 'EVENT', // dispatch action results in { type: 'EVENT' }
// function
dispatch: (action, state) => ({ type: `PREFIX_${action.type}` })
}Make an ajax request using axios library.
{
// object
request: {
method: 'get',
url: 'url',
cancelWhen: [
'IF_REQUEST_IS_PENDING_CANCEL_IT_WHEN_THIS_IS_DISPATCHED',
'OR_THIS'
],
onSuccess: 'DISPATCH_THIS_IF_AJAX_SUCCEDED'
onFail: 'DISPATCH_THIS_IF_AJAX_FAILED',
// other axios props
},
// function
request: (action, state) => { ... }
}For convenience aliases have been provided for all supported request methods:
{
post: { ... },
get: { ... },
del: { ... },
head: { ... },
options: { ... },
put: { ... }
patch: { ... }
}Dispatch event or make a request, after an action is debounced
{
// integer
debounce: 500, // in ms
// function
debounce: (action, state) => state.debounceConfig
}Dispatch event or make a request, after an action is delayed
{
// integer
delay: 500, // in ms
// function
delay: (action, state) => state.delayConfig
}If defined, no events will reach a reducer unless it's defined in a process manager.
{
validate: false // default
}