How to check if object is Immutable? - immutable.js

Immutable object can be an instance of:
Immutable.List
Immutable.Map
Immutable.OrderedMap
Immutable.Set
Immutable.OrderedSet
Immutable.Stack

There is an open ticket to improve the API which is on the roadmap for 4.0. Until this is implemented, I suggest you use Immutable.Iterable.isIterable() (docs).
Using instanceof is not reliable (e. g. returns false when different modules use different copies of Immutable.js)

I have learned that using instanceof to determine wether object is Immutable is unsafe:
Module A:
var Immutable = require('immutable');
module.exports = Immutable.Map({foo: "bar});
Module B:
var Immutable = require('immutable');
var moduleA = require('moduleA');
moduleA instanceof Immutable.Map // will return false
Immutable.js API defines the following methods to check if object is an instance of Immutable:
Map.isMap()
List.isList()
Stack.isStack()
OrderedMap.isOrderedMap()
Set.isSet()
OrderedSet.isOrderedSet()
and
Iterable.isIterable()
The latter checks if:
True if an Iterable, or any of its subclasses.
List, Stack, Map, OrderedMap, Set and OrderedSet are all subclasses of Iterable.

Immutable.js has isImmutable() function since v4.0.0-rc.1:
import { isImmutable, Map, List, Stack } from 'immutable';
isImmutable([]); // false
isImmutable({}); // false
isImmutable(Map()); // true
isImmutable(List()); // true
isImmutable(Stack()); // true
isImmutable(Map().asMutable()); // false
If you use one of the previous versions, you can check if object is Immutable this way:
Immutable.Iterable.isIterable(YOUR_ENTITY)
because all immutables inherit from the Iterable object

And this way you can get to know what type of Immutable Iterable variable is:
const obj0 = 'xxx';
const obj1 = Immutable.fromJS({x: 'XXX', z: 'ZZZ'});
const obj2 = Immutable.fromJS([ {x: 'XXX'}, {z: 'ZZZ'}]);
const types = ['List', 'Stack', 'Map', 'OrderedMap', 'Set', 'OrderedSet'];
const type0 = types.find(currType => Immutable[currType][`is${currType}`](obj0));
const type1 = types.find(currType => Immutable[currType][`is${currType}`](obj1));
const type2 = types.find(currType => Immutable[currType][`is${currType}`](obj2));
console.log(`Obj0 is: ${type0}`); // Obj0 is: undefined
console.log(`Obj1 is: ${type1}`); // Obj1 is: Map
console.log(`Obj2 is: ${type2}`); // Obj2 is: List
<script src="https://cdnjs.cloudflare.com/ajax/libs/immutable/3.8.1/immutable.js"></script>

Checking specific types will generally cause more work later on. Usually I would wait to lock types in by checking for Map or List, but...
My motivation here is mostly that my call .get of undefined poops itself really hard, and initializing properly all over the place helps, but doesn't catch all edge cases. I just want the data or undefined without any breakage. Specific type checking causes me to do more work later if I want it to make changes.
This looser version solves many more edge cases(most if not all extend type Iterable which has .get, and all data is eventually gotten) than a specific type check does(which usually only saves you when you try to update on the wrong type etc).
/* getValid: Checks for valid ImmutableJS type Iterable
returns valid Iterable, valid Iterable child data, or undefined
Iterable.isIterable(maybeIterable) && maybeIterable.get(['data', key], Map()), becomes
getValid(maybeIterable, ['data', key], Map())
But wait! There's more! As a result:
getValid(maybeIterable) returns the maybeIterable or undefined
and we can still say getValid(maybeIterable, null, Map()) returns the maybeIterable or Map() */
export const getValid = (maybeIterable, path, getInstead) =>
Iterable.isIterable(maybeIterable) && path
? ((typeof path === 'object' && maybeIterable.getIn(path, getInstead)) || maybeIterable.get(path, getInstead))
: Iterable.isIterable(maybeIterable) && maybeIterable || getInstead;
//Here is an untested version that a friend requested. It is slightly easier to grok.
export const getValid = (maybeIterable, path, getInstead) => {
if(valid(maybeIterable)) { // Check if it is valid
if(path) { // Check if it has a key
if(typeof path === 'object') { // Check if it is an 'array'
return maybeIterable.getIn(path, getInstead) // Get your stuff
} else {
maybeIterable.get(path, getInstead) // Get your stuff
}
} else {
return maybeIterable || getInstead; // No key? just return the valid Iterable
}
} else {
return undefined; // Not valid, return undefined, perhaps should return false here
}
}
Just give me what I am asking for or tell me no. Don't explode. I believe underscore does something similar also.

This may work in some cases:
typeof object.toJS === 'function'
You can use this ducktyping method if you check immutable vs plain objects (json), for example.

Related

Knex sometime return an empty string

I'm using knex with MYSQL. I have a function that I called to show the data to the user, Also I'm using a view table which has 5 right join on it and I think it will take some time to return values from the table plus I added the WHERE condition on my knex and it looks like this :
var showClass = (teacherId , ClassId){
return new Promise((resolve , reject)=>{
knex.select().from('v_cardex_details').where({teacherId }).andWhere({id : ClassId}).then(classes =>{
resolve(classes)
}).catch(err=>{
console.error(`Show Teacher class Error: ${err}`)
reject (err)
})
})
}
and I call this general function to response some request something like this
exports.EditClass = (req,res)=>{
knex('Table').update({//Some update stuff here}).then(()=>{
showClass(req.user.id, req.params.id).then(data=>{
return res.status(200).json({data , message:''})
})
}).catch()
}
With the same input, this function after updating returns value and some times it returns an empty string, especially when it's on the hosting server most of the time it returns nothing but { message : '' }
Try to create simplified code by removing all the unnecessary wrappers and you might find where your problem is. AFAIK there is no way that that your {data , message:''} would create an object containing just {message: ''} without any additional attributes.
> var data = []
undefined
> {data, foo:1}
{ data: [], foo: 1 }
> data = undefined
undefined
> {data, foo:1}
{ data: undefined, foo: 1 }
> {data1, foo:1}
ReferenceError: data1 is not defined
The problem you are experiencing does not exist in from the code you have shared (though there are syntax errors and other problems).
EDIT:
res.json() uses JSON.stringify() to convert js object to JSON strings. So if value of data in your code is undefined instead of and array, that could explain the behavior you are experiencing:
λ node
> JSON.stringify({ test: undefined })
'{}'
As you can see JSON.stringify() omits the attributes with value undefined from the output JSON string.

redux-saga takeLeading action plus additional parameter

I've implement a redux effect takeLeading that will ignore subsequent actions if the saga is currently running:
export const takeLeading = (patternOrChannel, saga, ...args) => fork(function*() {
while (true) {
const action = yield take(patternOrChannel);
yield call(saga, ...args.concat(action));
}
});
I use this for API fetching in my application, where each endpoint in my API has its own action type. So for GET methods it's useful to block if the request has already been dispatched somewhere else in the app. The saga looks like:
return function* () {
yield all([takeLeading(GET_USER_ID, callApiGen), takeLeading(GET_WIDGET_ID, callApiGen)]);
}
The obvious problem is that if I want to get two different user IDs, the second will block because it too has action type GET_USER_ID. Short of making a different action for each possible parameter, is there a way to implement some takeLeadingForFunc(<action>, (action) => <id>, saga) that allows me to keep the concise format of specifying one effect per request type but allows me to not block if the <id> is different? I was trying to wrap takeLeading with takeEvery to implement something but couldn't quite get it.
EDIT:
I got something like this to work:
export const takeLeadingForFunc = (f) => (patternOrChannel, saga, ...args) => fork(function*() {
let takeLeadings = {};
while (true) {
const action = yield take(patternOrChannel);
if (!(f(action) in takeLeadings)) {
yield call(saga, ...args.concat(action))
takeLeadings[f(action)] = yield takeLeading((ac) => f(ac) === f(action) && ac.type === action.type, saga, ...args)
}
}
});
Which takes an extractor function f that should return a primitive. This feels kind of hacky, so was wondering if there's a more idiomatic way to do this.

Multicast observable: attempting to subscribe results in "cannot read property 'subscribe' of undefined" error

I have a need in my code to perform an AJAX request and send the resulting data to two different places, so I figured using a multicast observable was the easiest way of achieving this. My code looks like this:
In the constructor for my 'app' object:
this.getEpisodeDescription = (id) => jsonLoader("http://www.randomtext.me/api/lorem/p-2/8-24", "text_out");
function jsonLoader (url, field)
{
let stream = Rx.Observable.ajax ({ url: url, crossDomain: true })
.retry (1)
.pluck ("response");
if (field !== undefined)
return stream.pluck(field);
else
return stream;
}
I've successfully used this method before to retrieve data for a single receiver, so I'm sure this is working OK. The caller is new, however:
loadSummary (id)
{
let cachedValue = this.summaries.get(id);
if (cachedValue !== undefined) return Rx.Observable.of(cachedValue);
let observable = this.app.getEpisodeDescription(id);
let multicast = observable.multicast ().refCount ();
multicast.subscribe(result => this.summaries.put(id, result));
return multicast;
}
When I try executing this method, I get the following stack trace:
Uncaught TypeError: Cannot read property 'subscribe' of undefined
at Observable.ConnectableObservable._subscribe (app.js:44193)
at Observable._trySubscribe (app.js:10253)
at Observable.subscribe (app.js:10241)
at RefCountOperator.call (app.js:44275)
at Observable.subscribe (app.js:10238)
at AsyncAction.SubscribeOnObservable.dispatch (app.js:71532)
at AsyncAction._execute (app.js:21083)
at AsyncAction.execute (app.js:21058)
at AsyncScheduler.flush (app.js:21156)
(Ignore file name and line numbers -- I'm using webpack and it doesn't seem to be producing a working line number map at the moment)
Any ideas what's going on? Specifically, how does it happen that I get an object out of the call to multicast that has appropriate subscribe etc methods, but when you try to subscribe to it it apparently can't subscribe to the parent?
The first parameter to the multicast() operator is either Subject factory function or a Subject instance.
This means you should be using it like this if you want to have one shared Subject instance:
let multicast = observable.multicast(new Subject()).refCount();
... or like this to make a new Subject for every observer:
let multicast = observable.multicast(() => new Subject()).refCount();

fromJS to nested ordered maps with Immutable.js?

does anyone have a suggestion for the best way to deep convert a js list of lists to nested ordered maps with Immutable.js?
You can create your own custom conversion. For example, to turn JS objects into Immutable.OrderedMap:
function fromJSOrdered(js) {
return typeof js !== 'object' || js === null ? js :
Array.isArray(js) ?
Immutable.Seq(js).map(fromJSOrdered).toList() :
Immutable.Seq(js).map(fromJSOrdered).toOrderedMap();
}
The fromJS has a second parameter called reviver, which can be exactly used for this.
import Immutable from 'immutable';
const reviver = (key, value) =>
Immutable.Iterable.isKeyed(value) ? value.toOrderedMap() : value.toList();
const data = Immutable.fromJS(js, reviver);
the response by #Albert Olivé had problems on my case use, because of recursion and lack of main context. I've tried with a second context argument to map, with problems again.
Finally I realized that I didn't care about order in submaps, just in main map passed to the function, to maintain order provided by server on html lists. So I changed the function to no recursive this way:
fromJSOrdered(js) {
if (typeof js !== 'object' || js === null || Array.isArray(js)) {
return fromJS(js);
}
return new OrderedMap(fromJS(js));
}
I actually published a package recently that can transform an object, array or Map of objects into Immutable List's and Record's:
https://github.com/jhukdev/immutable-parsejs
You can take a look at the source if you don't want Record's, easy to change.
Record's are nice though, as you have direct property access, meaning in future if you wanted to switch away from ImmutableJs, it's an easier prospect

How can I update an existing JSON object's parameters in Grails?

I'm making a todo list. When first entering the item and adding it to the list, the server works great. It takes the parameters that the user selects and passes them into a list on the server that can be viewed by rendering Item.list(), that looks like so:
[{"class":"server.Item","id":1,"assignedTo":"User 1","comments":null,"completed":false,"creator":"User 1","name":"Task 1","priority":"1","type":"Personal"},
{"class":"server.Item","id":2,"assignedTo":"User 2","comments":null,"completed":false,"creator":"User 2","name":"Er","priority":"3","type":"Work"},
{"class":"server.Item","id":3,"assignedTo":"User 1","comments":null,"completed":false,"creator":"User 2","name":"Ga","priority":"1","type":"Work"}]
Now, the user then has the option to edit the task later. On the client side this works fine, but then I need the user to be able to save the new, updated task.
This is my current update function:
def updateList() {
def newItem = Item.findById(request.JSON.id)
newItem.assignedTo = request.JSON.assignedTo
newItem.comments = request.JSON.comments
newItem.completed = request.JSON.completed
newItem.creator = request.JSON.creator
newItem.name = request.JSON.name
newItem.priority = request.JSON.priority
newItem.type = request.JSON.type
newItem.save(flush: true)
render newItem as JSON
}
This doesn't work, however. I get a null pointer exception that says "Cannot set property "assignedTo" on null object. I'm assuming that the findById request is not getting anything for the JSON object, and thus there is no object to assign values to, however I don't know what the problem is considering the items are in fact being put into the Item.list().
This is called with the following JS function on the client side:
$scope.updateList = function() {
angular.forEach($scope.items, function (item) {
// serverList.save({command: 'updateList'}, item);
$http.post('http://localhost:8080/server/todoList/updateList', item)
.success(function(response) {})
.error(function(response) {alert("Failed to update");});
});
};
This might depend on your Grails version, but you should be able to do this:
def update(Item item) {
if (!item) {
// return a 404
} else {
// you should really use a service and not save
// in the controller
itemService.update(item)
respond item
}
}
Grails is smart enough look that item up since there is an ID in the JSON params, and populate the object correctly.
Sort of a work around for anyone else that may need to do this in a basic manner, what I've done that works is clear the list when "Update List" is clicked, then read back in the values that are currently in the client side list.
Grails:
def clearList() {
Item.executeUpdate('delete from Item')
render Item.list()
}
def updateList() {
def newItem = new Item(request.JSON)
newItem.save(flush:true)
render newItem as JSON
}
Javascript:
$scope.updateList = function() { // Update list on the server
serverList.get({command: 'clearList'});
angular.forEach($scope.items, function (item) {
serverList.save({command: 'updateList'}, item);
});
};