So I'm building a basic facet search in React using data from an API, but I'm struggling with the adding & removing of values from the payload sent back to the server. The payload object can contain Arrays, Objects & Strings. So say I have a payload structure like the following:
payload = {
search_query: ""
topic: [],
genre: [],
rating: "",
type: {}
}
Using deepmerge I'm able to pass multiple values back to the API which is working fine, so an example payload would be...
payload = {
search_query: "Luther"
topic: ["9832748273", "4823794872394"],
genre: ["3827487483", "3287483274873"],
rating: "18",
type: {
args: {
when: "today",
min: 15
},
name: "spot"
}
}
So far so good, I get the expected results back. Now I have a toggle on the facet to remove it from the payload, which sends back the value to a function to remove from the payload. For example:
Clear Search, value to remove = "Luther"
Toggle of topic, value to remove = "9832748273"
Toggle of genre, value to remove = "3827487483"
Toggle of rating, value to remove = "18"
Toggle of type, value to remove = { args: { when: "today", min: 15}, name: "spot"}
Search & Rating would return empty strings, topic & genre would remove items from the arrays and type would return an empty object.
This works for removing the array values but feels dirty and I want a clean way to handle all types!
const removeObjValue = (obj, filterValue) => {
Object.entries(obj).forEach(([key, value]) => {
Object.entries(value).forEach(([subKey, subValue]) => {
if(subvalue === filterValue) {
if(Array.isArray(value)) {
const index = value.indexOf(subvalue);
if (index > -1) {
value.splice(index, 1);
}
}
}
});
});
return obj;
}
I just use delete keyword to remove object attributes, like this
if (payload[key]) {
if (payload[key] instanceof Array) {
var idx = payload[key].indexOf(value);
if (idx > -1) payload[key].splice(idx, 1);
} else {
delete payload[key];
}
}
code example
Related
If you have an array as part of your state, and that array contains objects, whats an easy way to update the state with a change to one of those objects?
Example, modified from the tutorial on react:
var CommentBox = React.createClass({
getInitialState: function() {
return {data: [
{ id: 1, author: "john", text: "foo" },
{ id: 2, author: "bob", text: "bar" }
]};
},
handleCommentEdit: function(id, text) {
var existingComment = this.state.data.filter({ function(c) { c.id == id; }).first();
var updatedComments = ??; // not sure how to do this
this.setState({data: updatedComments});
}
}
I quite like doing this with Object.assign rather than the immutability helpers.
handleCommentEdit: function(id, text) {
this.setState({
data: this.state.data.map(el => (el.id === id ? Object.assign({}, el, { text }) : el))
});
}
I just think this is much more succinct than splice and doesn't require knowing an index or explicitly handling the not found case.
If you are feeling all ES2018, you can also do this with spread instead of Object.assign
this.setState({
data: this.state.data.map(el => (el.id === id ? {...el, text} : el))
});
While updating state the key part is to treat it as if it is immutable. Any solution would work fine if you can guarantee it.
Here is my solution using immutability-helper:
jsFiddle:
var update = require('immutability-helper');
handleCommentEdit: function(id, text) {
var data = this.state.data;
var commentIndex = data.findIndex(function(c) {
return c.id == id;
});
var updatedComment = update(data[commentIndex], {text: {$set: text}});
var newData = update(data, {
$splice: [[commentIndex, 1, updatedComment]]
});
this.setState({data: newData});
},
Following questions about state arrays may also help:
Correct modification of state arrays in ReactJS
what is the preferred way to mutate a React state?
I'm trying to explain better how to do this AND what's going on.
First, find the index of the element you're replacing in the state array.
Second, update the element at that index
Third, call setState with the new collection
import update from 'immutability-helper';
// this.state = { employees: [{id: 1, name: 'Obama'}, {id: 2, name: 'Trump'}] }
updateEmployee(employee) {
const index = this.state.employees.findIndex((emp) => emp.id === employee.id);
const updatedEmployees = update(this.state.employees, {$splice: [[index, 1, employee]]}); // array.splice(start, deleteCount, item1)
this.setState({employees: updatedEmployees});
}
Edit: there's a much better way to do this w/o a 3rd party library
const index = this.state.employees.findIndex(emp => emp.id === employee.id);
employees = [...this.state.employees]; // important to create a copy, otherwise you'll modify state outside of setState call
employees[index] = employee;
this.setState({employees});
You can do this with multiple way, I am going to show you that I mostly used. When I am working with arrays in react usually I pass a custom attribute with current index value, in the example below I have passed data-index attribute, data- is html 5 convention.
Ex:
//handleChange method.
handleChange(e){
const {name, value} = e,
index = e.target.getAttribute('data-index'), //custom attribute value
updatedObj = Object.assign({}, this.state.arr[i],{[name]: value});
//update state value.
this.setState({
arr: [
...this.state.arr.slice(0, index),
updatedObj,
...this.state.arr.slice(index + 1)
]
})
}
MirageJS provides all model ids as strings. Our backend uses integers, which are convenient for sorting and so on. After reading around MirageJS does not support integer IDs out of the box. From the conversations I've read the best solution would be to convert Ids in a serializer.
Output:
{
id: "1",
title: "Some title",
otherValue: "Some other value"
}
But what I want is:
Expected Output:
{
id: 1,
title: "Some title",
otherValue: "Some other value"
}
I really want to convert ALL ids. This would included nested objects, and serialized Ids.
I think you should be able to use a custom IdentityManager for this. Here's a REPL example. (Note: REPL is a work in progress + currently only works on Chrome).
Here's the code:
import { Server, Model } from "miragejs";
class IntegerIDManager {
constructor() {
this.ids = new Set();
this.nextId = 1;
}
// Returns a new unused unique identifier.
fetch() {
let id = this.nextId++;
this.ids.add(id);
return id;
}
// Registers an identifier as used. Must throw if identifier is already used.
set(id) {
if (this.ids.has(id)) {
throw new Error('ID ' + id + 'has already been used.');
}
this.ids.add(id);
}
// Resets all used identifiers to unused.
reset() {
this.ids.clear();
}
}
export default new Server({
identityManagers: {
application: IntegerIDManager,
},
models: {
user: Model,
},
seeds(server) {
server.createList("user", 3);
},
routes() {
this.resource("user");
},
});
When I make a GET request to /users with this server I get integer IDs back.
My solution is to traverse the data and recursively convert all Ids. It's working pretty well.
I have a number of other requirements, like removing the data key and embedding or serializing Ids.
const ApplicationSerializer = Serializer.extend({
root: true,
serialize(resource, request) {
// required to serializedIds
// handle removing root key
const json = Serializer.prototype.serialize.apply(this, arguments)
const root = resource.models
? this.keyForCollection(resource.modelName)
: this.keyForModel(resource.modelName)
const keyedItem = json[root]
// convert single string id to integer
const idToInt = id => Number(id)
// convert array of ids to integers
const idsToInt = ids => ids.map(id => idToInt(id))
// check if the data being passed is a collection or model
const isCollection = data => Array.isArray(data)
// check if data should be traversed
const shouldTraverse = entry =>
Array.isArray(entry) || entry instanceof Object
// check if the entry is an id
const isIdKey = key => key === 'id'
// check for serialized Ids
// don't be stupid and create an array of values with a key like `arachnIds`
const isIdArray = (key, value) =>
key.slice(key.length - 3, key.length) === 'Ids' && Array.isArray(value)
// traverse the passed model and update Ids where required, keeping other entries as is
const traverseModel = model =>
Object.entries(model).reduce(
(a, c) =>
isIdKey(c[0])
? // convert id to int
{ ...a, [c[0]]: idToInt(c[1]) }
: // convert id array to int
isIdArray(c[0], c[1])
? { ...a, [c[0]]: idsToInt(c[1]) }
: // traverse nested entries
shouldTraverse(c[1])
? { ...a, [c[0]]: applyFuncToModels(c[1]) }
: // keep regular entries
{ ...a, [c[0]]: c[1] },
{}
)
// start traversal of data
const applyFuncToModels = data =>
isCollection(data)
? data.map(model =>
// confirm we're working with a model, and not a value
model instance of Object ? traverseModel(model) : model)
: traverseModel(data)
return applyFuncToModels(keyedItem)
}
})
I had to solve this problem as well (fingers crossed that this gets included into the library) and my use case is simpler than the first answer.
function convertIdsToNumbers(o) {
Object.keys(o).forEach((k) => {
const v = o[k]
if (Array.isArray(v) || v instanceof Object) convertIdsToNumbers(v)
if (k === 'id' || /.*Id$/.test(k)) {
o[k] = Number(v)
}
})
}
const ApplicationSerializer = RestSerializer.extend({
root: false,
embed: true,
serialize(object, request) {
let json = Serializer.prototype.serialize.apply(this, arguments)
convertIdsToNumbers(json)
return {
status: request.status,
payload: json,
}
},
})
So my call returns something like:
data:
{
nameData: 'Test33333',
emailData: email#email.com,
urlLink: link.com
additionalDetails: [
{
field: 'email',
value: 'other#email.com'
},
{
field: 'name',
value: 'name1223'
}
]
}
Now, I want to make a function that would take the passed parameter (data) and make an array of objects, that should look like below. It should be done in more generic way.
Array output expectation:
fullData = [
{
name: 'data_name'
value: 'Test33333'
},
{
name: 'data_email',
value: 'email#email.com'
},
{
name: 'data_url',
value: 'Link.com'
},
extraData: [
//we never know which one will it return
]
];
It should be done in the function, with name, for example:
generateDataFromObj(data)
so
generateDataArrFromObj = (data) => {
//logic here that will map correctly the data
}
How can this be achieved? I am not really proficient with JavaScript, thanks.
Assuming that you keep your data property keys in camelCase this will work for any data you add, not just the data in the example. Here I've used planetLink. It reduces over the object keys using an initial empty array), extracts the new key name from the existing property key, and concatenates each new object to the returned array.
const data = { nameData: 'Test33333', emailData: 'email#email.com', planetLink: 'Mars' };
function generateDataArrFromObj(data) {
const regex = /([a-z]+)[A-Z]/;
// `reduce` over the object keys
return Object.keys(data).reduce((acc, c) => {
// match against the lowercase part of the key value
// and create the new key name `data_x`
const key = `data_${c.match(regex)[1]}`;
return acc.concat({ name: key, value: data[c] });
}, []);
}
console.log(generateDataArrFromObj(data));
Just run a map over the object keys, this will return an array populated by each item, then in the func map runs over each item, build an object like so:
Object.keys(myObj).map(key => {return {name: key, value: myObj[key]}})
Given an immutable state like this:
alerts: {
5a8c76171bbb57b2950000c4: [
{
_id:5af7c8652552070000000064
device_id:5a8c76171bbb57b2950000c4
count: 1
},
{
_id:5af7c8722552070000000068
device_id:5a8c76171bbb57b2950000c4
count: 2
}
]
}
and an object like this:
{
_id:5af7c8652552070000000064
device_id:5a8c76171bbb57b2950000c4
count: 2
}
I want to replace the object with the same id in the alerts state (immutable), such that end result looks like this:
alerts: {
5a12356ws13tch: [
{
_id:5af7c8652552070000000064
device_id:5a8c76171bbb57b2950000c4
count: 2
},
{
_id:5af7c8722552070000000068
device_id:5a8c76171bbb57b2950000c4
count: 2
}
]
}
How can I do that? With mergeDeep, getIn, setIn, and updateIn, found on List, Map or OrderedMap ?
I tried doing something like this.. where index is 0 and deviceId is 5a12356ws13tch
Does not work though.
export const oneAlertFetched = (state, {deviceId, index, alert}) => state.setIn(['alerts', deviceId, index], alert).merge({fetching: false})
I tried this as well. Does not work.
export const oneAlertFetched = (state, {deviceId, index, alert}) => {
const a = state.alerts[deviceId][index]
state.alerts[deviceId][index] = Object.assign({}, a, alert)
return
}
By immutable, you mean that your property is non-writable.
If you want to modify your object in-place (not recommended), you will need the property to be at least configurable:
const device = alerts['5a12356ws13tch'][0];
if (Object.getOwnPropertyDescriptor(device, 'count').configurable) {
// Manually make it `writable`
Object.defineProperty(device, 'count', {
writable: true
});
// Update property's value
device.count++;
// Set it back to `non-writable`
Object.defineProperty(device, 'count', {
writable: false
});
}
console.log(device.count); // 2
If it is not configurable (cannot make it writable), or you do not want to jeopardize your application (it must be non-writable on purpose), then you should work on copies.
const device = alerts['5a12356ws13tch'][0];
alerts['5a12356ws13tch'][0] = Object.assign({}, device, {count: device.count + 1});
Object.assign() works on flat objects. If you need deep copy, have a look at my SO answer there.
I think you mean you want to return a new object with the updated payload?
function getNextAlerts(alerts, parentDeviceId, payload) {
const alertsForDevice = alerts[parentDeviceId];
if (!alertsForDevice || alertsForDevice.length === 0) {
console.log('No alerts for device', deviceId);
return;
}
return {
...alerts,
[parentDeviceId]: alerts[parentDeviceId].map(item =>
item._id === payload._id ? payload : item
),
}
}
const alerts = {
'5a12356ws13tch': [
{
_id: '5af7c8652552070000000064',
device_id: '5a8c76171bbb57b2950000c4',
count: 1
},
{
_id: '5af7c8722552070000000068',
device_id: '5a8c76171bbb57b2950000c4',
count: 2
}
]
};
const nextAlerts = getNextAlerts(alerts, '5a12356ws13tch', {
_id: '5af7c8652552070000000064',
device_id: '5a8c76171bbb57b2950000c4',
count: 2,
});
console.log('nextAlerts:', nextAlerts);
<script src="https://cdnjs.cloudflare.com/ajax/libs/lodash.js/4.17.10/lodash.min.js"></script>
If you're working with plain JavaScript objects and want to keep "immutable" approach you have to use spreads all over the nested structure of state object.
But, there are some tools already targeting this issue - lenses.
Here is the example of both approaches, array/object spreads and lenses - ramda repl.
In short, your example via spreads:
const oneAlertFetched = (state, { deviceId, index, alert }) => ({
...state,
alerts: {
...state.alerts,
[deviceId]: [
...state.alerts[deviceId].slice(0, index),
{ ...state.alerts[deviceId][index], ...alert },
...state.alerts[deviceId].slice(index + 1)
],
}
})
And via lenses using Ramda's over, lensPath, merge and __*:
const oneAlertFetched = (state, { deviceId, index, alert }) =>
R.over(
R.lensPath(['alerts', deviceId, index]),
R.merge(R.__, alert),
state
)
* R.__ placeholder used to swap 1st & 2nd parameters of R.merge
PS: lenses solution is intentionally adjusted to match the declaration of your function, so you can easily compare two approaches. However, in real life, with such powerful and flexible tool, we can rewrite the function to be more readable, reusable, and performant.
I want to compare the data.examples array object name.value property value with wcObject.notCoveredList key, If the key matches I want to push all matched values of wcObject to an array inorder to display in UI. If the key doesn't match, I want the name.desc property value of data.examples array object to be pushed by removing the not covered text at the end.
data = {
examples : [
{
name: {
value:"someOne",
desc: "some random word not covered"
},
type: {
value:"General",
desc:"General"
}
}, {
name: {
value:"secondOne",
desc: "second on in the queue not covered"
},
type: {
value:"General",
desc:"General"
}
}, {
name: {
value:"thirdOne",
desc: "third one from the last not covered"
},
type: {
value:"General",
desc:"General"
}
}
]
}
wcObject = {
notCoveredList : [
{ someOne: "anyone can start " },
{ secondOne: "second One cannot start" },
{ thirdOne: "third One can be allowed" }
]
}
So, this code:
Builds up a filter object. We grab all the keys of
wcObject.notCoveredList, and make them keys on a single object (with
an undefined value), so that we can look up those keys with a single
hasOwnProperty() call instead of iterating over an array when we need to filter.
Maps every member of the data.examples array to either its own name.desc property or [stripped of ' not covered'] name.value property.
.
wcNotCoveredKeys = wcObject.notCoveredList.reduce((memo, item) => {
// value is empty for all, we only care about keys.
memo[Object.keys(item)[0]] = undefined;
return memo;
}, {})
// having built up our lookup table of those not covered, we continue:
forUI = data.examples.map(example => {
if (wcNotCoveredKeys.hasOwnProperty(example.name.value)) {
return example.name.value;
}
else {
notCoveredString = example.name.desc;
cutOutIndex = notCoveredString.indexOf(' not covered');
return notCoveredString.slice(0, cutOutIndex)
}
});
(updated to integrate string slicing)
Just to be clear: if you removed the second item from wcObject.notCoveredList, then the output you'd get in forUI (given the example data structures/values you provided) would be
["someOne", "second on in the queue", "thirdOne"]