Hvordan jeg arkitekterede en React-applikation på en enkelt side

Med datastrukturer, komponenter og integration med Redux

Jeg har for nylig bygget en applikation med en side, der interagerer med en backend JSON API-server. Jeg valgte at bruge React til at uddybe min forståelse af React-grundlæggende og hvordan hvert værktøj kan hjælpe med at opbygge en skalerbar frontend.

Stakken til denne applikation består af:

  • Frontend med React / Redux
  • En backend JSON API-server med Sinatra, integreret med Postgres til databasepresistens
  • En API-klient, der henter data fra OMDb API, skrevet i Ruby

For dette indlæg antager vi, at vi har backend gennemført. Så lad os fokusere på, hvordan designbeslutninger træffes på frontend.

Sidebemærkning: Beslutningerne, der præsenteres her, er kun til reference og kan variere afhængigt af din ansøgnings behov. Et eksempel på OMDb Movie Tracker-app bruges her til demonstration.

Appen

Ansøgningen består af en søgeformular. En bruger kan indtaste en filmtitel for at returnere et filmresultat fra OMDb. Brugeren kan også gemme en film med en vurdering og en kort kommentar på en favoritliste.

Klik her for at se den endelige app. Klik her for at se kildekoden.

Når en bruger søger i en film på startsiden, ser det sådan ud:

Af hensyn til enkelheden fokuserer vi kun på at designe programmets kerneegenskaber i denne artikel. Du kan også springe til del II: Reduxaf serien.

Datastruktur

Definition af passende datastrukturer bør være et af de vigtigste aspekter ved design af en app. Dette skal komme som det første trin, da det ikke kun bestemmer, hvordan frontend skal gengive elementerne, men også hvordan API-serveren skal returnere JSON-svarene.

Til denne app skal vi bruge to hovedstykker information til korrekt gengivelse af vores brugergrænseflade: et enkelt filmresultat og en liste over foretrukne film .

Filmresultatobjekt

Et enkelt filmresultat indeholder information såsom titel, år, beskrivelse og plakatbillede. Med dette skal vi definere et objekt, der kan gemme disse attributter:

{ "title": "Star Wars: Episode IV - A New Hope", "year": "1977", "plot": "Luke Skywalker joins forces with a Jedi Knight...", "poster": "//m.media-amazon.com/path/to/poster.jpg", "imdbID": "tt0076759"}

Den posteregenskab er simpelthen en URL til plakat billede, der vil blive vist i resultaterne. Hvis der ikke er nogen plakat tilgængelig for den film, vil den være "N / A", som vi viser en pladsholder. Vi har også brug for en imdbIDattribut til entydigt at identificere hver film. Dette er nyttigt til at afgøre, om der allerede findes et filmresultat på favoritlisten. Vi undersøger senere, hvordan det fungerer.

Foretrukne liste

Favoritlisten indeholder alle de film, der er gemt som favoritter. Listen vil se sådan ud:

[ { title: "Star Wars", year: "1977", ..., rating: 4 }, { title: "Avatar", year: "2009", ..., rating: 5 }]

Husk, at vi bliver nødt til at slå en bestemt film op fra listen, og tidskompleksiteten for denne tilgang er O (N) . Mens det fungerer fint til mindre datasæt, så forestil dig at skulle søge efter en film på en favoritliste, der vokser på ubestemt tid.

Med dette i tankerne valgte jeg at gå med en hash-tabel med nøgler som imdbIDog værdier som foretrukne filmobjekter:

{ tt0076759: { title: "Star Wars: Episode IV - A New Hope", year: "1977", plot: "...", poster: "...", rating: "4", comment: "May the force be with you!", }, tt0499549: { title: "Avatar", year: "2009", plot: "...", poster: "...", rating: "5", comment: "Favorite movie!", }}

Med dette kan vi slå en film op på favoritlisten i O (1) tid efter dens imdbID.

Bemærk: runtime-kompleksiteten betyder sandsynligvis ikke noget i de fleste tilfælde, da datasættene normalt er små på klientsiden. Vi vil også udføre udskæring og kopiering (også O (N) -operationer) i Redux alligevel. Men som ingeniør er det godt at være opmærksom på potentielle optimeringer, som vi kan udføre.

Komponenter

Komponenter er kernen i React. Vi bliver nødt til at bestemme, hvilke der vil interagere med Redux-butikken, og hvilke der kun er til præsentation. Vi kan også genbruge nogle af præsentationskomponenterne. Vores komponenthierarki vil se sådan ud:

Forside

Vi udpeger vores App- komponent på øverste niveau. Når rodstien er besøgt, skal den gengive SearchContainer . Det skal også vise flashbeskeder til brugeren og håndtere routing på klientsiden.

Den SearchContainer vil hente filmen resultat fra vores Redux butik, give oplysninger som rekvisitter til MovieItem til destruktion. Det sender også en søgning, når en bruger indsender en søgning i SearchInputForm . Mere om Redux senere.

Føj til favoritformular

Når brugeren klikker på knappen "Føj til favoritter", viser vi AddFavoriteForm , en kontrolleret komponent.

Vi opdaterer konstant dens tilstand, hver gang en bruger ændrer klassificeringen eller indtaster teksten i kommentartextområdet. Dette er nyttigt til validering ved indsendelse af formular.

Den RatingForm er ansvarlig for at gengive de gule stjerner, når brugeren klikker på dem. Den informerer også den aktuelle vurderingsværdi til AddFavoriteForm .

Fanen Favoritter

Når en bruger klikker på fanen "Foretrukne", gengiver appen FavoritesContainer .

Den FavoritesContainer er ansvarlig for at hente favoritlisten fra Redux butikken. Det sender også handlinger, når en bruger ændrer en vurdering eller klikker på knappen "Fjern".

Our MovieItem and FavoritesInfo are simply presentational components that receive props from FavoritesContainer.

We’ll reuse the RatingForm component here. When a user clicks on a star in the RatingForm, the FavoritesContainer receives the rating value and dispatches an update rating action to the Redux store.

Redux Store

Our Redux store will include reducers that handle the search and favorites actions. Additionally, we’ll need to include a status reducer to track state changes when a user initiates an action. We’ll explore more on the status reducer later.

//store.js
import { createStore, combineReducers, applyMiddleware } from 'redux';import thunk from "redux-thunk";
import search from './reducers/searchReducer';import favorites from './reducers/favoritesReducer';import status from './reducers/statusReducer';
export default createStore( combineReducers({ search, favorites, status }), {}, applyMiddleware(thunk))

We’ll also apply the Redux Thunk middleware right away. We’ll go more into detail on that later. Now, let’s figure out how we manage the state changes when a user submits a search.

Search Reducer

When a user performs a search action, we want to update the store with a new search result via searchReducer. We can then render our components accordingly. The general flow of events looks like this:

We’ll treat “Get search result” as a black box for now. We’ll explore how that works later with Redux Thunk. Now, let’s implement the reducer function.

//searchReducer.js
const initialState = { "title": "", "year": "", "plot": "", "poster": "", "imdbID": "",}
export default (state = initialState, action) => { if (action.type === 'SEARCH_SUCCESS') { state = action.result; } return state;}

The initialState will represent the data structure defined earlier as a single movie result object. In the reducer function, we handle the action where a search is successful. If the action is triggered, we simply reassign the state to the new movie result object.

//searchActions.jsexport const searchSuccess = (result) => ({ type: 'SEARCH_SUCCESS', result});

We define an action called searchSuccess that takes in a single argument, the movie result object, and returns an action object of type “SEARCH_SUCCESS”. We will dispatch this action upon a successful search API call.

Redux Thunk: Search

Let’s explore how the “Get search result” from earlier works. First, we need to make a remote API call to our backend API server. When the request receives a successful JSON response, we’ll dispatch the searchSuccess action along with the payload to searchReducer.

Knowing that we’ll need to dispatch after an asynchronous call completes, we’ll make use of Redux Thunk. Thunk comes into play for making multiple dispatches or delaying a dispatch. With Thunk, our updated flow of events looks like this:

For this, we define a function that takes in a single argument title and serves as the initial search action. Thisfunction is responsible for fetching the search result and dispatching a searchSuccess action:

//searchActions.jsimport apiClient from '../apiClient';
...
export function search(title) { return (dispatch) => { apiClient.query(title) .then(response => { dispatch(searchSuccess(response.data)) }); }}

We’ve set up our API client beforehand, and you can read more about how I set up the API client here. The apiClient.query method simply performs an AJAX GET request to our backend server and returns a Promise with the response data.

We can then connect this function as an action dispatch to our SearchContainer component:

//SearchContainer.js
import React from 'react';import { connect } from 'react-redux';import { search } from '../actions/searchActions';
...
const mapStateToProps = (state) => ( { result: state.search, });
const mapDispatchToProps = (dispatch) => ( { search(title) { dispatch(search(title)) }, });
export default connect(mapStateToProps, mapDispatchToProps)(SearchContainer);

When a search request succeeds, our SearchContainer component will render the movie result:

Handling Other Search Statuses

Now we have our search action working properly and connected to our SearchContainer component, we’d like to handle other cases other than a successful search.

Search request pending

When a user submits a search, we’ll display a loading animation to indicate that the search request is pending:

Search request succeeds

If the search fails, we’ll display an appropriate error message to the user. This is useful to provide some context. A search failure could happen in cases where a movie title is not available, or our server is experiencing issues communicating with the OMDb API.

To handle different search statuses, we’ll need a way to store and update the current status along with any error messages.

Status Reducer

The statusReducer is responsible for tracking state changes whenever a user performs an action. The current state of an action can be represented by one of the three “statuses”:

  • Pending (when a user first initiates the action)
  • Success (when a request returns a successful response)
  • Error (when a request returns an error response)

With these statuses in place, we can render different UIs based on the current status of a given action type. In this case, we’ll focus on tracking the status of the search action.

We’ll start by implementing the statusReducer. For the initial state, we need to track the current search status and any errors:

// statusReducer.jsconst initialState = { search: '', // status of the current search searchError: '', // error message when a search fails}

Next, we need to define the reducer function. Whenever our SearchContainer dispatches a “SEARCH_[STATUS]” action, we will update the store by replacing the search and searchError properties.

// statusReducer.js
...
export default (state = initialState, action) => { const actionHandlers = { 'SEARCH_REQUEST': { search: 'PENDING', searchError: '', }, 'SEARCH_SUCCESS': { search: 'SUCCESS', searchError: '', }, 'SEARCH_FAILURE': { search: 'ERROR', searchError: action.error, }, } const propsToUpdate = actionHandlers[action.type]; state = Object.assign({}, state, propsToUpdate); return state;}

We use an actionHandlers hash table here since we are only replacing the state’s properties. Furthermore, it improves readability more than using if/else or case statements.

With our statusReducer in place, we can render the UI based on different search statuses. We will update our flow of events to this:

We now have additional searchRequest and searchFailure actions available to dispatch to the store:

//searchActions.js
export const searchRequest = () => ({ type: 'SEARCH_REQUEST'});
export const searchFailure = (error) => ({ type: 'SEARCH_FAILURE', error});

To update our search action, we will dispatch searchRequest immediately and will dispatch searchSuccess or searchFailure based on the eventual success or failure of the Promise returned by Axios:

//searchActions.js
...
export function search(title) { return (dispatch) => { dispatch(searchRequest());
apiClient.query(title) .then(response => { dispatch(searchSuccess(response.data)) }) .catch(error => { dispatch(searchFailure(error.response.data)) }); }}

We can now connect the search status state to our SearchContainer, passing it as a prop. Whenever our store receives the state changes, our SearchContainer renders a loading animation, an error message, or the search result:

//SearchContainer.js
...(imports omitted)
const SearchContainer = (props) => (   props.search(title) } /> { (props.searchStatus === 'SUCCESS') ?  : null } { (props.searchStatus === 'PENDING') ?    : null } { (props.searchStatus === 'ERROR') ?  

{ props.searchError }

: null } );
const mapStateToProps = (state) => ( { searchStatus: state.status.search, searchError: state.status.searchError, result: state.search, });
...

Favorites Reducer

We’ll need to handle CRUD actions performed by a user on the favorites list. Recalling from our API endpoints earlier, we’d like to allow users to perform the following actions and update our store accordingly:

  • Save a movie into the favorites list
  • Retrieve all favorited movies
  • Update a favorite’s rating
  • Delete a movie from the favorites list

To ensure that the reducer function is pure, we simply copy the old state into a new object together with any new properties usingObject.assign. Note that we only handle actions with types of _SUCCESS:

//favoritesReducer.js
export default (state = {}, action) => { switch (action.type) { case 'SAVE_FAVORITE_SUCCESS': state = Object.assign({}, state, action.favorite); break;
case 'GET_FAVORITES_SUCCESS': state = action.favorites; break;
case 'UPDATE_RATING_SUCCESS': state = Object.assign({}, state, action.favorite); break;
case 'DELETE_FAVORITE_SUCCESS': state = Object.assign({}, state); delete state[action.imdbID]; break;
default: return state; } return state;}

We’ll leave the initialState as an empty object. The reason is that if our initialState contains placeholder movie items, our app will render them immediately before waiting for the actual favorites list response from our backend API server.

From now on, each of the favorites action will follow a general flow of events illustrated below. The pattern is similar to the search action in the previous section, except right now we’ll skip handling any “PENDING” status.

Save Favorites Action

Take the save favorites action for example. The function makes an API call to with our apiClient and dispatches either a saveFavoriteSuccess or a saveFavoriteFailure action, depending on whether or not we receive a successful response:

//favoritesActions.jsimport apiClient from '../apiClient';
export const saveFavoriteSuccess = (favorite) => ({ type: 'SAVE_FAVORITE_SUCCESS', favorite});
export const saveFavoriteFailure = (error) => ({ type: 'SAVE_FAVORITE_FAILURE', error});
export function save(movie) { return (dispatch) => { apiClient.saveFavorite(movie) .then(res => { dispatch(saveFavoriteSuccess(res.data)) }) .catch(err => { dispatch(saveFavoriteFailure(err.response.data)) }); }}

We can now connect the save favorite action to AddFavoriteForm through React Redux.

To read more about how I handled the flow to display flash messages, click here.

Conclusion

Designing the frontend of an application requires some forethought, even when using a popular JavaScript library such as React. By thinking about how the data structures, components, APIs, and state management work as a whole, we can better anticipate edge cases and effectively fix errors when they arise. By using certain design patterns such as controlled components, Redux, and handling AJAX workflow using Thunk, we can streamline managing the flow of providing UI feedback to user actions. Ultimately, how we approach the design will have an impact on usability, clarity, and future scalability.

References

Fullstack React: The Complete Guide to ReactJS and Friends

About me

Jeg er softwareingeniør i NYC og medskaber af SpaceCraft. Jeg har erfaring med at designe applikationer på en side, synkronisere tilstand mellem flere klienter og implementere skalerbare applikationer med Docker.

Jeg leder i øjeblikket efter min næste fuldtidsmulighed! Kontakt os, hvis du tror, ​​at jeg passer godt til dit team.