登陆和授权是小程序开发会遇到的第一个问题，这里把相关业务逻辑、工具代码抽取出来，展示我们如何引入的一些包使得原生微信小程序内也可以使用 async/await、fetch、localStorage、状态管理、GraphQL 等等特性，希望对大家有所帮助。 前端 目录结构 [代码]├── app.js ├── app.json ├── app.wxss ├── common │ └── api │ └── index.js ├── config.js ├── pages │ └── index │ ├── api │ │ └── index.js │ ├── img │ │ ├── btn.png │ │ └── bg.jpg │ ├── index.js │ ├── index.json │ ├── index.wxml │ └── index.wxss ├── project.config.json ├── store │ ├── action.js │ └── index.js ├── utils │ └── index.js └── vendor ├── event-emitter.js ├── fetch.js ├── fetchql.js ├── http.js ├── promisify.js ├── regenerator.js ├── storage.js └── store.js [代码] 业务代码 app.js [代码]import store from './store/index' const { loginInfo } = store.state App({ store, onLaunch() { // 打开小程序即登陆，无需用户授权可获得 openID if(!loginInfo) store.dispatch('login') }, }) [代码] store/index.js [代码]import Store from '../vendor/store' import localStorage from '../vendor/storage' import actions from './action' const loginInfo = localStorage.getItem('loginInfo') export default new Store({ state: { // 在全局状态中维护登陆信息 loginInfo, }, actions, }) [代码] store/action.js [代码]import regeneratorRuntime from '../vendor/regenerator'; import wx from '../vendor/promisify'; import localStorage from '../vendor/storage' import api from '../common/api/index'; export default { async login({ state }, payload) { const { code } = await wx.loginAsync(); const { authSetting } = await wx.getSettingAsync() // 如果用户曾授权，直接可以拿到 encryptedData const { encryptedData, iv } = authSetting['scope.userInfo'] ? await wx.getUserInfoAsync({ withCredentials: true }) : {}; // 如果用户未曾授权，也可以拿到 openID const { token, userInfo } = await api.login({ code, encryptedData, iv }); // 为接口统一配置 Token getApp().gql.requestObject.headers['Authorization'] = `Bearer ${token}`; // 本地缓存登陆信息 localStorage.setItem('loginInfo', { token, userInfo } ) return { loginInfo: { token, userInfo } } } } [代码] common/api/index.js [代码]import regeneratorRuntime from '../../vendor/regenerator.js' export default { /** * 登录接口 * 如果只有 code，只返回 token，如果有 encryptedData, iv，同时返回用户的昵称和头像 * @param {*} param0 */ async login({ code, encryptedData, iv }) { const query = `query login($code: String!, $encryptedData: String, $iv: String){ login(code:$code, encryptedData:$encryptedData, iv:$iv, appid:$appid){ token userInfo { nickName avatarUrl } } }` const { login: { token, userInfo } } = await getApp().query({ query, variables: { code, encryptedData, iv } }) return { token, userInfo } }, } [代码] pages/index/index.js [代码]import regeneratorRuntime from '../../vendor/regenerator.js' const app = getApp() Page({ data: {}, onLoad(options) { // 将用户登录信息注入到当前页面的 data 中，并且当数据在全局范围内被更新时，都会自动刷新本页面 app.store.mapState(['loginInfo'], this) }, async login({ detail: { errMsg } }) { if (errMsg === 'getUserInfo:fail auth deny') return app.store.dispatch('login') // 继续处理业务 }, }) [代码] pages/index/index.wxml [代码]<view class="container"> <form report-submit="true" bindsubmit="saveFormId"> <button form-type="submit" open-type="getUserInfo" bindgetuserinfo="login">登录</button> </form> </view> [代码] 工具代码 事件处理 vendor/event-emitter.js [代码]const id_Identifier = '__id__'; function randomId() { return Math.random().toString(36).substr(2, 16); } function findIndexById(id) { return this.findIndex(item => item[id_Identifier] === id); } export default class EventEmitter { constructor() { this.events = {} } /** * listen on a event * @param event * @param listener */ on(event, listener) { let { events } = this; let container = events[event] || []; let id = randomId(); let index; listener[id_Identifier] = id; container.push(listener); return () => { index = findIndexById.call(container, id); index >= 0 && container.splice(index, 1); } }; /** * remove all listen of an event * @param event */ off (event) { this.events[event] = []; }; /** * clear all event listen */ clear () { this.events = {}; }; /** * listen on a event once, if it been trigger, it will cancel the listner * @param event * @param listener */ once (event, listener) { let { events } = this; let container = events[event] || []; let id = randomId(); let index; let callback = () => { index = findIndexById.call(container, id); index >= 0 && container.splice(index, 1); listener.apply(this, arguments); }; callback[id_Identifier] = id; container.push(callback); }; /** * emit event */ emit () { const { events } = this; const argv = [].slice.call(arguments); const event = argv.shift(); ((events['*'] || []).concat(events[event] || [])).map(listener => self.emitting(event, argv, listener)); }; /** * define emitting * @param event * @param dataArray * @param listener */ emitting (event, dataArray, listener) { listener.apply(this, dataArray); }; } [代码] 封装 wx.request() 接口 vendor/http.js [代码]import EventEmitter from './event-emitter.js'; const DEFAULT_CONFIG = { maxConcurrent: 10, timeout: 0, header: {}, dataType: 'json' }; class Http extends EventEmitter { constructor(config = DEFAULT_CONFIG) { super(); this.config = config; this.ctx = wx; this.queue = []; this.runningTask = 0; this.maxConcurrent = DEFAULT_CONFIG.maxConcurrent; this.maxConcurrent = config.maxConcurrent; this.requestInterceptor = () => true; this.responseInterceptor = () => true; } create(config = DEFAULT_CONFIG) { return new Http(config); } next() { const queue = this.queue; if (!queue.length || this.runningTask >= this.maxConcurrent) return; const entity = queue.shift(); const config = entity.config; const { requestInterceptor, responseInterceptor } = this; if (requestInterceptor.call(this, config) !== true) { let response = { data: null, errMsg: `Request Interceptor: Request can\'t pass the Interceptor`, statusCode: 0, header: {} }; entity.reject(response); return; } this.emit('request', config); this.runningTask = this.runningTask + 1; let timer = null; let aborted = false; let finished = false; const callBack = { success: (res) => { if (aborted) return; finished = true; timer && clearTimeout(timer); entity.response = res; this.emit('success', config, res); responseInterceptor.call(this, config, res) !== true ? entity.reject(res) : entity.resolve(res); }, fail: (res) => { if (aborted) return; finished = true; timer && clearTimeout(timer); entity.response = res; this.emit('fail', config, res); responseInterceptor.call(this, config, res) !== true ? entity.reject(res) : entity.resolve(res); }, complete: () => { if (aborted) return; this.emit('complete', config, entity.response); this.next(); this.runningTask = this.runningTask - 1; } }; const requestConfig = Object.assign(config, callBack); const task = this.ctx.request(requestConfig); if (this.config.timeout > 0) { timer = setTimeout(() => { if (!finished) { aborted = true; task && task.abort(); this.next(); } }, this.config.timeout); } } request(method, url, data, header, dataType = 'json') { const config = { method, url, data, header: { ...header, ...this.config.header }, dataType: dataType || this.config.dataType }; return new Promise((resolve, reject) => { const entity = { config, resolve, reject, response: null }; this.queue.push(entity); this.next(); }); } head(url, data, header, dataType) { return this.request('HEAD', url, data, header, dataType); } options(url, data, header, dataType) { return this.request('OPTIONS', url, data, header, dataType); } get(url, data, header, dataType) { return this.request('GET', url, data, header, dataType); } post(url, data, header, dataType) { return this.request('POST', url, data, header, dataType); } put(url, data, header, dataType) { return this.request('PUT', url, data, header, dataType); } ['delete'](url, data, header, dataType) { return this.request('DELETE', url, data, header, dataType); } trace(url, data, header, dataType) { return this.request('TRACE', url, data, header, dataType); } connect(url, data, header, dataType) { return this.request('CONNECT', url, data, header, dataType); } setRequestInterceptor(interceptor) { this.requestInterceptor = interceptor; return this; } setResponseInterceptor(interceptor) { this.responseInterceptor = interceptor; return this; } clean() { this.queue = []; } } export default new Http(); [代码] 兼容 fetch 标准 vendor/fetch.js [代码]import http from './http'; const httpClient = http.create({ maxConcurrent: 10, timeout: 0, header: {}, dataType: 'json' }); function generateResponse(res) { let header = res.header || {}; let config = res.config || {}; return { ok: ((res.statusCode / 200) | 0) === 1, // 200-299 status: res.statusCode, statusText: res.errMsg, url: config.url, clone: () => generateResponse(res), text: () => Promise.resolve( typeof res.data === 'string' ? res.data : JSON.stringify(res.data) ), json: () => { if (typeof res.data === 'object') return Promise.resolve(res.data); let json = {}; try { json = JSON.parse(res.data); } catch (err) { console.error(err); } return json; }, blob: () => Promise.resolve(new Blob([res.data])), headers: { keys: () => Object.keys(header), entries: () => { let all = []; for (let key in header) { if (header.hasOwnProperty(key)) { all.push([key, header[key]]); } } return all; }, get: n => header[n.toLowerCase()], has: n => n.toLowerCase() in header } }; } export default (typeof fetch === 'function' ? fetch.bind() : function(url, options) { options = options || {}; return httpClient .request(options.method || 'get', url, options.body, options.headers) .then(res => Promise.resolve(generateResponse(res))) .catch(res => Promise.reject(generateResponse(res))); }); [代码] GraphQL客户端 vendor/fetchql.js [代码]import fetch from './fetch'; // https://github.com/gucheen/fetchql /** Class to realize fetch interceptors */ class FetchInterceptor { constructor() { this.interceptors = []; /* global fetch */ this.fetch = (...args) => this.interceptorWrapper(fetch, ...args); } /** * add new interceptors * @param {(Object|Object[])} interceptors */ addInterceptors(interceptors) { const removeIndex = []; if (Array.isArray(interceptors)) { interceptors.map((interceptor) => { removeIndex.push(this.interceptors.length); return this.interceptors.push(interceptor); }); } else if (interceptors instanceof Object) { removeIndex.push(this.interceptors.length); this.interceptors.push(interceptors); } this.updateInterceptors(); return () => this.removeInterceptors(removeIndex); } /** * remove interceptors by indexes * @param {number[]} indexes */ removeInterceptors(indexes) { if (Array.isArray(indexes)) { indexes.map(index => this.interceptors.splice(index, 1)); this.updateInterceptors(); } } /** * @private */ updateInterceptors() { this.reversedInterceptors = this.interceptors .reduce((array, interceptor) => [interceptor].concat(array), []); } /** * remove all interceptors */ clearInterceptors() { this.interceptors = []; this.updateInterceptors(); } /** * @private */ interceptorWrapper(fetch, ...args) { let promise = Promise.resolve(args); this.reversedInterceptors.forEach(({ request, requestError }) => { if (request || requestError) { promise = promise.then(() => request(...args), requestError); } }); promise = promise.then(() => fetch(...args)); this.reversedInterceptors.forEach(({ response, responseError }) => { if (response || responseError) { promise = promise.then(response, responseError); } }); return promise; } } /** * GraphQL client with fetch api. * @extends FetchInterceptor */ class FetchQL extends FetchInterceptor { /** * Create a FetchQL instance. * @param {Object} options * @param {String} options.url - the server address of GraphQL * @param {(Object|Object[])=} options.interceptors * @param {{}=} options.headers - request headers * @param {FetchQL~requestQueueChanged=} options.onStart - callback function of a new request queue * @param {FetchQL~requestQueueChanged=} options.onEnd - callback function of request queue finished * @param {Boolean=} options.omitEmptyVariables - remove null props(null or '') from the variables * @param {Object=} options.requestOptions - addition options to fetch request(refer to fetch api) */ constructor({ url, interceptors, headers, onStart, onEnd, omitEmptyVariables = false, requestOptions = {}, }) { super(); this.requestObject = Object.assign( {}, { method: 'POST', headers: Object.assign({}, { Accept: 'application/json', 'Content-Type': 'application/json', }, headers), credentials: 'same-origin', }, requestOptions, ); this.url = url; this.omitEmptyVariables = omitEmptyVariables; // marker for request queue this.requestQueueLength = 0; // using for caching enums' type this.EnumMap = {}; this.callbacks = { onStart, onEnd, }; this.addInterceptors(interceptors); } /** * operate a query * @param {Object} options * @param {String} options.operationName * @param {String} options.query * @param {Object=} options.variables * @param {Object=} options.opts - addition options(will not be passed to server) * @param {Boolean=} options.opts.omitEmptyVariables - remove null props(null or '') from the variables * @param {Object=} options.requestOptions - addition options to fetch request(refer to fetch api) * @returns {Promise} * @memberOf FetchQL */ query({ operationName, query, variables, opts = {}, requestOptions = {}, }) { const options = Object.assign({}, this.requestObject, requestOptions); let vars; if (this.omitEmptyVariables || opts.omitEmptyVariables) { vars = this.doOmitEmptyVariables(variables); } else { vars = variables; } const body = { operationName, query, variables: vars, }; options.body = JSON.stringify(body); this.onStart(); return this.fetch(this.url, options) .then((res) => { if (res.ok) { return res.json(); } // return an custom error stack if request error return { errors: [{ message: res.statusText, stack: res, }], }; }) .then(({ data, errors }) => ( new Promise((resolve, reject) => { this.onEnd(); // if data in response is 'null' if (!data) { return reject(errors || [{}]); } // if all properties of data is 'null' const allDataKeyEmpty = Object.keys(data).every(key => !data[key]); if (allDataKeyEmpty) { return reject(errors); } return resolve({ data, errors }); }) )); } /** * get current server address * @returns {String} * @memberOf FetchQL */ getUrl() { return this.url; } /** * setting a new server address * @param {String} url * @memberOf FetchQL */ setUrl(url) { this.url = url; } /** * get information of enum type * @param {String[]} EnumNameList - array of enums' name * @returns {Promise} * @memberOf FetchQL */ getEnumTypes(EnumNameList) { const fullData = {}; // check cache status const unCachedEnumList = EnumNameList.filter((element) => { if (this.EnumMap[element]) { // enum has been cached fullData[element] = this.EnumMap[element]; return false; } return true; }); // immediately return the data if all enums have been cached if (!unCachedEnumList.length) { return new Promise((resolve) => { resolve({ data: fullData }); }); } // build query string for uncached enums const EnumTypeQuery = unCachedEnumList.map(type => ( `${type}: __type(name: "${type}") { ...EnumFragment }` )); const query = ` query { ${EnumTypeQuery.join('\n')} } fragment EnumFragment on __Type { kind description enumValues { name description } }`; const options = Object.assign({}, this.requestObject); options.body = JSON.stringify({ query }); this.onStart(); return this.fetch(this.url, options) .then((res) => { if (res.ok) { return res.json(); } // return an custom error stack if request error return { errors: [{ message: res.statusText, stack: res, }], }; }) .then(({ data, errors }) => ( new Promise((resolve, reject) => { this.onEnd(); // if data in response is 'null' and have any errors if (!data) { return reject(errors || [{ message: 'Do not get any data.' }]); } // if all properties of data is 'null' const allDataKeyEmpty = Object.keys(data).every(key => !data[key]); if (allDataKeyEmpty && errors && errors.length) { return reject(errors); } // merge enums' data const passData = Object.assign(fullData, data); // cache new enums' data Object.keys(data).map((key) => { this.EnumMap[key] = data[key]; return key; }); return resolve({ data: passData, errors }); }) )); } /** * calling on a request starting * if the request belong to a new queue, call the 'onStart' method */ onStart() { this.requestQueueLength++; if (this.requestQueueLength > 1 || !this.callbacks.onStart) { return; } this.callbacks.onStart(this.requestQueueLength); } /** * calling on a request ending * if current queue finished, calling the 'onEnd' method */ onEnd() { this.requestQueueLength--; if (this.requestQueueLength || !this.callbacks.onEnd) { return; } this.callbacks.onEnd(this.requestQueueLength); } /** * Callback of requests queue changes.(e.g. new queue or queue finished) * @callback FetchQL~requestQueueChanged * @param {number} queueLength - length of current request queue */ /** * remove empty props(null or '') from object * @param {Object} input * @returns {Object} * @memberOf FetchQL * @private */ doOmitEmptyVariables(input) { const nonEmptyObj = {}; Object.keys(input).map(key => { const value = input[key]; if ((typeof value === 'string' && value.length === 0) || value === null || value === undefined) { return key; } else if (value instanceof Object) { nonEmptyObj[key] = this.doOmitEmptyVariables(value); } else { nonEmptyObj[key] = value; } return key; }); return nonEmptyObj; } } export default FetchQL; [代码] 将wx的异步接口封装成Promise vendor/promisify.js [代码]function promisify(wx) { let wxx = { ...wx }; for (let attr in wxx) { if (!wxx.hasOwnProperty(attr) || typeof wxx[attr] != 'function') continue; // skip over the sync method if (/sync$/i.test(attr)) continue; wxx[attr + 'Async'] = function asyncFunction(argv = {}) { return new Promise(function (resolve, reject) { wxx[attr].call(wxx, { ...argv, ...{ success: res => resolve(res), fail: err => reject(err) } }); }); }; } return wxx; } export default promisify(typeof wx === 'object' ? wx : {}); [代码] localstorage vendor/storage.js [代码]class Storage { constructor(wx) { this.wx = wx; } static get timestamp() { return new Date() / 1000; } static __isExpired(entity) { if (!entity) return true; return Storage.timestamp - (entity.timestamp + entity.expiration) >= 0; } static get __info() { let info = {}; try { info = this.wx.getStorageInfoSync() || info; } catch (err) { console.error(err); } return info; } setItem(key, value, expiration) { const entity = { timestamp: Storage.timestamp, expiration, key, value }; this.wx.setStorageSync(key, JSON.stringify(entity)); return this; } getItem(key) { let entity; try { entity = this.wx.getStorageSync(key); if (entity) { entity = JSON.parse(entity); } else { return null; } } catch (err) { console.error(err); return null; } // 没有设置过期时间, 则直接返回值 if (!entity.expiration) return entity.value; // 已过期 if (Storage.__isExpired(entity)) { this.remove(key); return null; } else { return entity.value; } } removeItem(key) { try { this.wx.removeStorageSync(key); } catch (err) { console.error(err); } return this; } clear() { try { this.wx.clearStorageSync(); } catch (err) { console.error(err); } return this; } get info() { let info = {}; try { info = this.wx.getStorageInfoSync(); } catch (err) { console.error(err); } return info || {}; } get length() { return (this.info.keys || []).length; } } export default new Storage(wx); [代码] 状态管理 vendor/store.js [代码]module.exports = class Store { constructor({ state, actions }) { this.state = state || {} this.actions = actions || {} this.ctxs = [] } // 派发action, 统一返回promise action可以直接返回state dispatch(type, payload) { const update = res => { if (typeof res !== 'object') return this.setState(res) this.ctxs.map(ctx => ctx.setData(res)) return res } if (typeof this.actions[type] !== 'function') return const res = this.actions[type](this, payload) return res.constructor.toString().match(/function\s*([^(]*)/)[1] === 'Promise' ? res.then(update) : new Promise(resolve => resolve(update(res))) } // 修改state的方法 setState(data) { this.state = { ...this.state, ...data } } // 根据keys获取state getState(keys) { return keys.reduce((acc, key) => ({ ...acc, ...{ [key]: this.state[key] } }), {}) } // 映射state到实例中，可在onload或onshow中调用 mapState(keys, ctx) { if (!ctx || typeof ctx.setData !== 'function') return ctx.setData(this.getState(keys)) this.ctxs.push(ctx) } } [代码] 兼容 async/await vendor/regenerator.js [代码]/** * Copyright (c) 2014-present, Facebook, Inc. * * This source code is licensed under the MIT license found in the * LICENSE file in the root directory of this source tree. */ var regeneratorRuntime = (function (exports) { "use strict"; var Op = Object.prototype; var hasOwn = Op.hasOwnProperty; var undefined; // More compressible than void 0. var $Symbol = typeof Symbol === "function" ? Symbol : {}; var iteratorSymbol = $Symbol.iterator || "@@iterator"; var asyncIteratorSymbol = $Symbol.asyncIterator || "@@asyncIterator"; var toStringTagSymbol = $Symbol.toStringTag || "@@toStringTag"; function wrap(innerFn, outerFn, self, tryLocsList) { // If outerFn provided and outerFn.prototype is a Generator, then outerFn.prototype instanceof Generator. var protoGenerator = outerFn && outerFn.prototype instanceof Generator ? outerFn : Generator; var generator = Object.create(protoGenerator.prototype); var context = new Context(tryLocsList || []); // The ._invoke method unifies the implementations of the .next, // .throw, and .return methods. generator._invoke = makeInvokeMethod(innerFn, self, context); return generator; } exports.wrap = wrap; // Try/catch helper to minimize deoptimizations. Returns a completion // record like context.tryEntries[i].completion. This interface could // have been (and was previously) designed to take a closure to be // invoked without arguments, but in all the cases we care about we // already have an existing method we want to call, so there's no need // to create a new function object. We can even get away with assuming // the method takes exactly one argument, since that happens to be true // in every case, so we don't have to touch the arguments object. The // only additional allocation required is the completion record, which // has a stable shape and so hopefully should be cheap to allocate. function tryCatch(fn, obj, arg) { try { return { type: "normal", arg: fn.call(obj, arg) }; } catch (err) { return { type: "throw", arg: err }; } } var GenStateSuspendedStart = "suspendedStart"; var GenStateSuspendedYield = "suspendedYield"; var GenStateExecuting = "executing"; var GenStateCompleted = "completed"; // Returning this object from the innerFn has the same effect as // breaking out of the dispatch switch statement. var ContinueSentinel = {}; // Dummy constructor functions that we use as the .constructor and // .constructor.prototype properties for functions that return Generator // objects. For full spec compliance, you may wish to configure your // minifier not to mangle the names of these two functions. function Generator() {} function GeneratorFunction() {} function GeneratorFunctionPrototype() {} // This is a polyfill for %IteratorPrototype% for environments that // don't natively support it. var IteratorPrototype = {}; IteratorPrototype[iteratorSymbol] = function () { return this; }; var getProto = Object.getPrototypeOf; var NativeIteratorPrototype = getProto && getProto(getProto(values([]))); if (NativeIteratorPrototype && NativeIteratorPrototype !== Op && hasOwn.call(NativeIteratorPrototype, iteratorSymbol)) { // This environment has a native %IteratorPrototype%; use it instead // of the polyfill. IteratorPrototype = NativeIteratorPrototype; } var Gp = GeneratorFunctionPrototype.prototype = Generator.prototype = Object.create(IteratorPrototype); GeneratorFunction.prototype = Gp.constructor = GeneratorFunctionPrototype; GeneratorFunctionPrototype.constructor = GeneratorFunction; GeneratorFunctionPrototype[toStringTagSymbol] = GeneratorFunction.displayName = "GeneratorFunction"; // Helper for defining the .next, .throw, and .return methods of the // Iterator interface in terms of a single ._invoke method. function defineIteratorMethods(prototype) { ["next", "throw", "return"].forEach(function(method) { prototype[method] = function(arg) { return this._invoke(method, arg); }; }); } exports.isGeneratorFunction = function(genFun) { var ctor = typeof genFun === "function" && genFun.constructor; return ctor ? ctor === GeneratorFunction || // For the native GeneratorFunction constructor, the best we can // do is to check its .name property. (ctor.displayName || ctor.name) === "GeneratorFunction" : false; }; exports.mark = function(genFun) { if (Object.setPrototypeOf) { Object.setPrototypeOf(genFun, GeneratorFunctionPrototype); } else { genFun.__proto__ = GeneratorFunctionPrototype; if (!(toStringTagSymbol in genFun)) { genFun[toStringTagSymbol] = "GeneratorFunction"; } } genFun.prototype = Object.create(Gp); return genFun; }; // Within the body of any async function, `await x` is transformed to // `yield regeneratorRuntime.awrap(x)`, so that the runtime can test // `hasOwn.call(value, "__await")` to determine if the yielded value is // meant to be awaited. exports.awrap = function(arg) { return { __await: arg }; }; function AsyncIterator(generator) { function invoke(method, arg, resolve, reject) { var record = tryCatch(generator[method], generator, arg); if (record.type === "throw") { reject(record.arg); } else { var result = record.arg; var value = result.value; if (value && typeof value === "object" && hasOwn.call(value, "__await")) { return Promise.resolve(value.__await).then(function(value) { invoke("next", value, resolve, reject); }, function(err) { invoke("throw", err, resolve, reject); }); } return Promise.resolve(value).then(function(unwrapped) { // When a yielded Promise is resolved, its final value becomes // the .value of the Promise<{value,done}> result for the // current iteration. result.value = unwrapped; resolve(result); }, function(error) { // If a rejected Promise was yielded, throw the rejection back // into the async generator function so it can be handled there. return invoke("throw", error, resolve, reject); }); } } var previousPromise; function enqueue(method, arg) { function callInvokeWithMethodAndArg() { return new Promise(function(resolve, reject) { invoke(method, arg, resolve, reject); }); } return previousPromise = // If enqueue has been called before, then we want to wait until // all previous Promises have been resolved before calling invoke, // so that results are always delivered in the correct order. If // enqueue has not been called before, then it is important to // call invoke immediately, without waiting on a callback to fire, // so that the async generator function has the opportunity to do // any necessary setup in a predictable way. This predictability // is why the Promise constructor synchronously invokes its // executor callback, and why async functions synchronously // execute code before the first await. Since we implement simple // async functions in terms of async generators, it is especially // important to get this right, even though it requires care. previousPromise ? previousPromise.then( callInvokeWithMethodAndArg, // Avoid propagating failures to Promises returned by later // invocations of the iterator. callInvokeWithMethodAndArg ) : callInvokeWithMethodAndArg(); } // Define the unified helper method that is used to implement .next, // .throw, and .return (see defineIteratorMethods). this._invoke = enqueue; } defineIteratorMethods(AsyncIterator.prototype); AsyncIterator.prototype[asyncIteratorSymbol] = function () { return this; }; exports.AsyncIterator = AsyncIterator; // Note that simple async functions are implemented on top of // AsyncIterator objects; they just return a Promise for the value of // the final result produced by the iterator. exports.async = function(innerFn, outerFn, self, tryLocsList) { var iter = new AsyncIterator( wrap(innerFn, outerFn, self, tryLocsList) ); return exports.isGeneratorFunction(outerFn) ? iter // If outerFn is a generator, return the full iterator. : iter.next().then(function(result) { return result.done ? result.value : iter.next(); }); }; function makeInvokeMethod(innerFn, self, context) { var state = GenStateSuspendedStart; return function invoke(method, arg) { if (state === GenStateExecuting) { throw new Error("Generator is already running"); } if (state === GenStateCompleted) { if (method === "throw") { throw arg; } // Be forgiving, per 25.3.3.3.3 of the spec: // https://people.mozilla.org/~jorendorff/es6-draft.html#sec-generatorresume return doneResult(); } context.method = method; context.arg = arg; while (true) { var delegate = context.delegate; if (delegate) { var delegateResult = maybeInvokeDelegate(delegate, context); if (delegateResult) { if (delegateResult === ContinueSentinel) continue; return delegateResult; } } if (context.method === "next") { // Setting context._sent for legacy support of Babel's // function.sent implementation. context.sent = context._sent = context.arg; } else if (context.method === "throw") { if (state === GenStateSuspendedStart) { state = GenStateCompleted; throw context.arg; } context.dispatchException(context.arg); } else if (context.method === "return") { context.abrupt("return", context.arg); } state = GenStateExecuting; var record = tryCatch(innerFn, self, context); if (record.type === "normal") { // If an exception is thrown from innerFn, we leave state === // GenStateExecuting and loop back for another invocation. state = context.done ? GenStateCompleted : GenStateSuspendedYield; if (record.arg === ContinueSentinel) { continue; } return { value: record.arg, done: context.done }; } else if (record.type === "throw") { state = GenStateCompleted; // Dispatch the exception by looping back around to the // context.dispatchException(context.arg) call above. context.method = "throw"; context.arg = record.arg; } } }; } // Call delegate.iterator[context.method](context.arg) and handle the // result, either by returning a { value, done } result from the // delegate iterator, or by modifying context.method and context.arg, // setting context.delegate to null, and returning the ContinueSentinel. function maybeInvokeDelegate(delegate, context) { var method = delegate.iterator[context.method]; if (method === undefined) { // A .throw or .return when the delegate iterator has no .throw // method always terminates the yield* loop. context.delegate = null; if (context.method === "throw") { // Note: ["return"] must be used for ES3 parsing compatibility. if (delegate.iterator["return"]) { // If the delegate iterator has a return method, give it a // chance to clean up. context.method = "return"; context.arg = undefined; maybeInvokeDelegate(delegate, context); if (context.method === "throw") { // If maybeInvokeDelegate(context) changed context.method from // "return" to "throw", let that override the TypeError below. return ContinueSentinel; } } context.method = "throw"; context.arg = new TypeError( "The iterator does not provide a 'throw' method"); } return ContinueSentinel; } var record = tryCatch(method, delegate.iterator, context.arg); if (record.type === "throw") { context.method = "throw"; context.arg = record.arg; context.delegate = null; return ContinueSentinel; } var info = record.arg; if (! info) { context.method = "throw"; context.arg = new TypeError("iterator result is not an object"); context.delegate = null; return ContinueSentinel; } if (info.done) { // Assign the result of the finished delegate to the temporary // variable specified by delegate.resultName (see delegateYield). context[delegate.resultName] = info.value; // Resume execution at the desired location (see delegateYield). context.next = delegate.nextLoc; // If context.method was "throw" but the delegate handled the // exception, let the outer generator proceed normally. If // context.method was "next", forget context.arg since it has been // "consumed" by the delegate iterator. If context.method was // "return", allow the original .return call to continue in the // outer generator. if (context.method !== "return") { context.method = "next"; context.arg = undefined; } } else { // Re-yield the result returned by the delegate method. return info; } // The delegate iterator is finished, so forget it and continue with // the outer generator. context.delegate = null; return ContinueSentinel; } // Define Generator.prototype.{next,throw,return} in terms of the // unified ._invoke helper method. defineIteratorMethods(Gp); Gp[toStringTagSymbol] = "Generator"; // A Generator should always return itself as the iterator object when the // @@iterator function is called on it. Some browsers' implementations of the // iterator prototype chain incorrectly implement this, causing the Generator // object to not be returned from this call. This ensures that doesn't happen. // See https://github.com/facebook/regenerator/issues/274 for more details. Gp[iteratorSymbol] = function() { return this; }; Gp.toString = function() { return "[object Generator]"; }; function pushTryEntry(locs) { var entry = { tryLoc: locs[0] }; if (1 in locs) { entry.catchLoc = locs[1]; } if (2 in locs) { entry.finallyLoc = locs[2]; entry.afterLoc = locs[3]; } this.tryEntries.push(entry); } function resetTryEntry(entry) { var record = entry.completion || {}; record.type = "normal"; delete record.arg; entry.completion = record; } function Context(tryLocsList) { // The root entry object (effectively a try statement without a catch // or a finally block) gives us a place to store values thrown from // locations where there is no enclosing try statement. this.tryEntries = [{ tryLoc: "root" }]; tryLocsList.forEach(pushTryEntry, this); this.reset(true); } exports.keys = function(object) { var keys = []; for (var key in object) { keys.push(key); } keys.reverse(); // Rather than returning an object with a next method, we keep // things simple and return the next function itself. return function next() { while (keys.length) { var key = keys.pop(); if (key in object) { next.value = key; next.done = false; return next; } } // To avoid creating an additional object, we just hang the .value // and .done properties off the next function object itself. This // also ensures that the minifier will not anonymize the function. next.done = true; return next; }; }; function values(iterable) { if (iterable) { var iteratorMethod = iterable[iteratorSymbol]; if (iteratorMethod) { return iteratorMethod.call(iterable); } if (typeof iterable.next === "function") { return iterable; } if (!isNaN(iterable.length)) { var i = -1, next = function next() { while (++i < iterable.length) { if (hasOwn.call(iterable, i)) { next.value = iterable[i]; next.done = false; return next; } } next.value = undefined; next.done = true; return next; }; return next.next = next; } } // Return an iterator with no values. return { next: doneResult }; } exports.values = values; function doneResult() { return { value: undefined, done: true }; } Context.prototype = { constructor: Context, reset: function(skipTempReset) { this.prev = 0; this.next = 0; // Resetting context._sent for legacy support of Babel's // function.sent implementation. this.sent = this._sent = undefined; this.done = false; this.delegate = null; this.method = "next"; this.arg = undefined; this.tryEntries.forEach(resetTryEntry); if (!skipTempReset) { for (var name in this) { // Not sure about the optimal order of these conditions: if (name.charAt(0) === "t" && hasOwn.call(this, name) && !isNaN(+name.slice(1))) { this[name] = undefined; } } } }, stop: function() { this.done = true; var rootEntry = this.tryEntries[0]; var rootRecord = rootEntry.completion; if (rootRecord.type === "throw") { throw rootRecord.arg; } return this.rval; }, dispatchException: function(exception) { if (this.done) { throw exception; } var context = this; function handle(loc, caught) { record.type = "throw"; record.arg = exception; context.next = loc; if (caught) { // If the dispatched exception was caught by a catch block, // then let that catch block handle the exception normally. context.method = "next"; context.arg = undefined; } return !! caught; } for (var i = this.tryEntries.length - 1; i >= 0; --i) { var entry = this.tryEntries[i]; var record = entry.completion; if (entry.tryLoc === "root") { // Exception thrown outside of any try block that could handle // it, so set the completion value of the entire function to // throw the exception. return handle("end"); } if (entry.tryLoc <= this.prev) { var hasCatch = hasOwn.call(entry, "catchLoc"); var hasFinally = hasOwn.call(entry, "finallyLoc"); if (hasCatch && hasFinally) { if (this.prev < entry.catchLoc) { return handle(entry.catchLoc, true); } else if (this.prev < entry.finallyLoc) { return handle(entry.finallyLoc); } } else if (hasCatch) { if (this.prev < entry.catchLoc) { return handle(entry.catchLoc, true); } } else if (hasFinally) { if (this.prev < entry.finallyLoc) { return handle(entry.finallyLoc); } } else { throw new Error("try statement without catch or finally"); } } } }, abrupt: function(type, arg) { for (var i = this.tryEntries.length - 1; i >= 0; --i) { var entry = this.tryEntries[i]; if (entry.tryLoc <= this.prev && hasOwn.call(entry, "finallyLoc") && this.prev < entry.finallyLoc) { var finallyEntry = entry; break; } } if (finallyEntry && (type === "break" || type === "continue") && finallyEntry.tryLoc <= arg && arg <= finallyEntry.finallyLoc) { // Ignore the finally entry if control is not jumping to a // location outside the try/catch block. finallyEntry = null; } var record = finallyEntry ? finallyEntry.completion : {}; record.type = type; record.arg = arg; if (finallyEntry) { this.method = "next"; this.next = finallyEntry.finallyLoc; return ContinueSentinel; } return this.complete(record); }, complete: function(record, afterLoc) { if (record.type === "throw") { throw record.arg; } if (record.type === "break" || record.type === "continue") { this.next = record.arg; } else if (record.type === "return") { this.rval = this.arg = record.arg; this.method = "return"; this.next = "end"; } else if (record.type === "normal" && afterLoc) { this.next = afterLoc; } return ContinueSentinel; }, finish: function(finallyLoc) { for (var i = this.tryEntries.length - 1; i >= 0; --i) { var entry = this.tryEntries[i]; if (entry.finallyLoc === finallyLoc) { this.complete(entry.completion, entry.afterLoc); resetTryEntry(entry); return ContinueSentinel; } } }, "catch": function(tryLoc) { for (var i = this.tryEntries.length - 1; i >= 0; --i) { var entry = this.tryEntries[i]; if (entry.tryLoc === tryLoc) { var record = entry.completion; if (record.type === "throw") { var thrown = record.arg; resetTryEntry(entry); } return thrown; } } // The context.catch method must only be called with a location // argument that corresponds to a known catch block. throw new Error("illegal catch attempt"); }, delegateYield: function(iterable, resultName, nextLoc) { this.delegate = { iterator: values(iterable), resultName: resultName, nextLoc: nextLoc }; if (this.method === "next") { // Deliberately forget the last sent value so that we don't // accidentally pass it on to the delegate. this.arg = undefined; } return ContinueSentinel; } }; // Regardless of whether this script is executing as a CommonJS module // or not, return the runtime object so that we can declare the variable // regeneratorRuntime in the outer scope, which allows this module to be // injected easily by `bin/regenerator --include-runtime script.js`. return exports; }( // If this script is executing as a CommonJS module, use module.exports // as the regeneratorRuntime namespace. Otherwise create a new empty // object. Either way, the resulting object will be used to initialize // the regeneratorRuntime variable at the top of this file. typeof module === "object" ? module.exports : {} )); [代码] 后端 [代码]const typeDefs = gql` # schema 下面是根类型，约定是 RootQuery 和 RootMutation schema { query: Query } # 定义具体的 Query 的结构 type Query { # 登陆接口 login(code: String!, encryptedData: String, iv: String): Login } type Login { token: String! userInfo: UserInfo } type UserInfo { nickName: String gender: String avatarUrl: String } `; const resolvers = { Query: { async login(parent, { code, encryptedData, iv }) { const { sessionKey, openId, unionId } = await wxService.code2Session(code); const userInfo = encryptedData && iv ? wxService.decryptData(sessionKey, encryptedData, iv) : { openId, unionId }; if (userInfo.nickName) { userService.createOrUpdateWxUser(userInfo); } const token = await userService.generateJwtToken(userInfo); return { token, userInfo }; }, }, }; [代码]

前言 在阿里和腾讯工作了6年，当了3年的前端面试官，把阿里和腾讯常问的面试题与答案汇总在我的Github中。希望对大家有所帮助，助力大家进入自己理想的企业。 项目地址是：https://github.com/airuikun/Weekly-FE-Interview 如果你在阿里和腾讯面试的时候遇到了什么不懂的问题，欢迎给我提issue，我会把答案和考点都列出来，公布在下一期的面试周刊里。 面试题精选 大家如果去阿里和腾讯面试过，就会发现，在网上刷了很多的前端面试题，但是去大厂面试的时候还是一头雾水，那是因为那些在网上一搜就能搜出来的题，大厂的面试官基本看不上，他们都会问一些开放题，在回答开放题的过程中，就能摸清你知识技能的广度和深度，所以本期会加入几道我在面试候选人常用的开放题，供大家学习和思考。 我把下面每道题的难度高低，和对标了阿里和腾讯的多少职级，都写上去了，大家可以参考一下自己是什么职级。 第 1 题：如何劫持https的请求，提供思路 难度：阿里p6+ ~ p7、腾讯t23 ~ t31 很多人在google上搜索“前端面试 + https详解”，把答案倒背如流，但是问到如何劫持https请求的时候就一脸懵逼，是因为还是停留在https理论性阶段。 想告诉大家的是，就算是https，也不是绝对的安全，以下提供一个本地劫持https请求的简单思路。 模拟中间人攻击，以百度为例 先用OpenSSL查看下证书，直接调用openssl库识别目标服务器支持的SSL/TLS cipher suite [代码] openssl s_client -connect www.baidu.com:443 [代码] 用sslcan识别ssl配置错误，过期协议，过时cipher suite和hash算法 [代码] sslscan -tlsall www.baidu.com:443 [代码] 分析证书详细数据 [代码] sslscan -show-certificate --no-ciphersuites www.baidu.com:443 [代码] 生成一个证书 [代码] openssl req -new -x509 -days 1096 -key ca.key -out ca.crt [代码] 开启路由功能 [代码] sysctl -w net.ipv4.ip_forward=1 [代码] 写转发规则，将80、443端口进行转发给8080和8443端口 [代码] iptables -t nat -A PREROUTING -p tcp --dport 80 -j REDIRECT --to-ports 8080 iptables -t nat -A PREROUTING -p tcp --dport 443 -j REDIRECT --to-ports 8443 [代码] 最后使用arpspoof进行arp欺骗 如果你有更好的想法或疑问，欢迎在这题目对应的github下留言：https://github.com/airuikun/Weekly-FE-Interview/issues/11 第 2 题：前端如何进行seo优化 难度：阿里p5、腾讯t21 合理的title、description、keywords：搜索对着三项的权重逐个减小，title值强调重点即可；description把页面内容高度概括，不可过分堆砌关键词；keywords列举出重要关键词。 语义化的HTML代码，符合W3C规范：语义化代码让搜索引擎容易理解网页 重要内容HTML代码放在最前：搜索引擎抓取HTML顺序是从上到下，保证重要内容一定会被抓取 重要内容不要用js输出：爬虫不会执行js获取内容 少用iframe：搜索引擎不会抓取iframe中的内容 非装饰性图片必须加alt 提高网站速度：网站速度是搜索引擎排序的一个重要指标 如果你有更好的答案或想法，欢迎在这题目对应的github下留言：https://github.com/airuikun/Weekly-FE-Interview/issues/12 第 3 题：前后端分离的项目如何seo 难度：阿里p6 ~ p6+、腾讯t22 ~ t23 使用prerender。但是回答prerender，面试官肯定会问你，如果不用prerender，让你直接去实现，好的，请看下面的第二个答案。 先去 https://www.baidu.com/robots.txt 找出常见的爬虫，然后在nginx上判断来访问页面用户的User-Agent是否是爬虫，如果是爬虫，就用nginx方向代理到我们自己用nodejs + puppeteer实现的爬虫服务器上，然后用你的爬虫服务器爬自己的前后端分离的前端项目页面，增加扒页面的接收延时，保证异步渲染的接口数据返回，最后得到了页面的数据，返还给来访问的爬虫即可。 如果你有更好的答案或想法，欢迎在这题目对应的github下留言：https://github.com/airuikun/Weekly-FE-Interview/issues/13 第 4 题：简单实现async/await中的async函数 难度：阿里p6 ~ p6+、腾讯t22 ~ t23 async 函数的实现原理，就是将 Generator 函数和自动执行器，包装在一个函数里 [代码]function spawn(genF) { return new Promise(function(resolve, reject) { const gen = genF(); function step(nextF) { let next; try { next = nextF(); } catch (e) { return reject(e); } if (next.done) { return resolve(next.value); } Promise.resolve(next.value).then( function(v) { step(function() { return gen.next(v); }); }, function(e) { step(function() { return gen.throw(e); }); } ); } step(function() { return gen.next(undefined); }); }); } [代码] 如果你有更好的答案或想法，欢迎在这题目对应的github下留言：https://github.com/airuikun/Weekly-FE-Interview/issues/14 第 5 题：1000-div问题 难度：阿里p5 ~ p6、腾讯t21 ~ t22 一次性插入1000个div，如何优化插入的性能 使用Fragment [代码] var fragment = document.createDocumentFragment(); fragment.appendChild(elem); [代码] 向1000个并排的div元素中，插入一个平级的div元素，如何优化插入的性能 先display:none 然后插入 再display:block 赋予key，然后使用virtual-dom，先render，然后diff，最后patch 脱离文档流，用GPU去渲染，开启硬件加速 如果你有更好的答案或想法，欢迎在这题目对应的github下留言：https://github.com/airuikun/Weekly-FE-Interview/issues/15 第 6 题：（开放题）2万小球问题：在浏览器端，用js存储2万个小球的信息，包含小球的大小，位置，颜色等，如何做到对这2万条小球信息进行最优检索和存储 难度：阿里p7、腾讯t31 你面试阿里和腾讯，能否上p7和t31，就看你对开放题能答有多深和多广。 这题目考察你如何在浏览器端中进行大数据的存储优化和检索优化。 如果你仅仅只是答用数组对象存储了2万个小球信息，然后用for循环去遍历进行索引，那是远远不够的。 这题要往深一点走，用特殊的数据结构和算法进行存储和索引。 然后进行存储和速度的一个权衡和对比，最终给出你认为的最优解。 我提供几个能触及阿里p7和腾讯t31级别的思路： 用ArrayBuffer实现极致存储 哈夫曼编码 + 字典查询树实现更优索引 用bit-map实现大数据筛查 用hash索引实现简单快捷的检索 用IndexedDB实现动态存储扩充浏览器端虚拟容量 用iframe的漏洞实现浏览器端localStorage无限存储，实现2千万小球信息存储 这种开放题答案不唯一，也不会要你现场手敲代码去实现，但是思路一定要行得通，并且是能打动面试官的思路，如果大家有更好的idea，欢迎大家到我的github里补充：https://github.com/airuikun/Weekly-FE-Interview/issues/16 第 7 题：（开放题）接上一题如何尽可能流畅的实现这2万小球在浏览器中，以直线运动的动效显示出来 难度：阿里p6+ ~ p7、腾讯t23 ~ t31 这题考察对大数据的动画显示优化，当然方法有很多种。 但是你有没有用到浏览器的高级api？ 你还有没有用到浏览器的专门针对动画的引擎？ 或者你对3D的实践和优化，都可以给面试官展示出来。 提供几个思路： 使用GPU硬件加速 使用webGL 使用assembly辅助计算，然后在浏览器端控制动画帧频 用web worker实现javascript多线程，分块处理小球 用单链表树算法和携程机制，实现任务动态分割和任务暂停、恢复、回滚，动态渲染和处理小球 如果大家有更好的idea，欢迎大家到我的github里补充：https://github.com/airuikun/Weekly-FE-Interview/issues/17 第 8 题：（开放题）100亿排序问题：内存不足，一次只允许你装载和操作1亿条数据，如何对100亿条数据进行排序。 难度：阿里p6+ ~ p7、腾讯t23 ~ t31 这题是考察算法和实际问题结合的一个问题 众所周知，腾讯玩的是社交，用户量极大。很多场景的数据量都是百亿甚至千亿级别。 那么如何对这些数据进行高效的操作呢，可以通过这题考察出来。 以前老听说很多人问，前端学算法没有用，考算法都是垃圾，面不出候选人的能力 其实。。。老哥实话告诉你，当你在做前端需要用到crc32、并查集、字典树、哈夫曼编码、LZ77之类东西的时候 已经是涉及到框架实现和极致优化层面了 那时你就已经到了另外一个前端高阶境界了 所以不要抵触算法，可能只是我们目前的眼界和能力，还没触及到那个层级 我前面已经公布了两道开放题的答案了，相信大家已经有所参悟。我觉得在思考开放题的过程中，会有很多意想不到的成长，所以我建议这道题大家可以尝试自己思考一下。本题答案会在周五公布到我的github上。 对应的github地址为：https://github.com/airuikun/Weekly-FE-Interview/issues/18 第 9 题：（开放题）a.b.c.d和a[‘b’][‘c’][‘d’]，哪个性能更高 难度：阿里p7 ~ p7+、腾讯t31 ~ t32 别看这题，题目上每个字都能看懂，但是里面涉及到的知识，暗藏杀鸡 这题要往深处走，会涉及ast抽象语法树、编译原理、v8内核对原生js实现问题 直接对标阿里p7 ~ p7+和腾讯t31 ~ t32职级，我觉得这个题是这篇文章里最难的一道题，所以我放在了开放题中的最后一题 大家多多思考，本题答案会在周五公布到我的github上 对应的github地址为：https://github.com/airuikun/Weekly-FE-Interview/issues/19 第 10 题：git时光机问题 难度：阿里p5 ~ p6+、腾讯t21 ~ t23 现在大厂，已经全部都是用git了，基本没人使用svn了 很多面试候选人对git只会commit、pull、push 但是有没有使用过reflog、cherry-pick等等，这些都很能体现出来你对代码管理的灵活程度和代码质量管理。 针对git时光机经典问题，我专门写了一个文章，轻松搞笑通俗易懂，大家可以看一下，放松放松，同时也能学到对git的时光机操作《git时光机》 结语 本人还写了一些前端进阶知识的文章，如果觉得不错可以点个star。 blog项目地址是：https://github.com/airuikun/blog 我是小蝌蚪，腾讯高级前端工程师，跟着我一起每周攻克几个前端技术难点。希望在小伙伴前端进阶的路上有所帮助，助力大家进入自己理想的企业。 交流 欢迎关注我的微信公众号，微信扫下面二维码或搜索“前端屌丝”，讲述了一个前端屌丝逆袭的心路历程，共勉。 [图片]

前言 最近一年多一直在做前端的一些测试，从小程序到店铺装修，基本都是纯前端的工作，刚开始从后端测试转为前端测试的时候，对前端东西茫然无感，而且团队内没有人做过纯前端的测试工作，只能一边踩坑一边总结经验，然后将容易出现问题的点形成体系、不断总结摸索，最终形成了目前的一套前端测试解决方案。在此，将有赞的前端质量保障体系进行总结，希望和大家一起交流。 先来全局看下有赞前端的技术架构和针对每个不同的层次，主要做了哪些保障质量的事情： [图片] [图片] 有赞的 Node 技术架构分为业务层、基础框架层、通用组件和基础服务层，我们日常比较关注的是基础框架、通用组件和业务层代码。Node 业务层做了两件事情，一是提供页面渲染的 client 层，用于和 C 端用户交互，包括样式、行为 js 等；二是提供数据服务的 server 层，用于组装后台提供的各种接口，完成面向 C 端的接口封装。 对于每个不同的层，我们都做了一些事情来保障质量，包括： 针对整个业务层的 UI 自动化、核心接口|页面拨测； 针对 client 层的 sentry 报警； 针对 server 层的接口测试、业务报警； 针对基础框架和通用组件的单元测试； 针对通用组件变更的版本变更报警； 针对线上发布的流程规范、用例维护等。 下面就来分别讲一下这几个维度的质量保障工作。 一、UI 自动化 很多人会认为，UI 自动化维护成本高、性价比低，但是为什么在有赞的前端质量保证体系中放在了最前面呢？ 前端重用户交互，单纯的接口测试、单元测试不能真实反映用户的操作路径，并且从以往的经验中总结得出，因为各种不可控因素导致的发布 A 功能而 B 功能无法使用，特别是核心简单场景的不可用时有出现，所以每次发布一个应用前，都会将此应用提供的核心功能执行一遍，那随着业务的不断积累，需要回归的测试场景也越来越多，导致回归的工作量巨大。为了降低人力成本，我们亟需通过自动化手段释放劳动力，所以将核心流程回归的 UI 自动化提到了最核心地位。 当然，UI 自动化的最大痛点确实是维护成本，为降低维护成本，我们将页面分为组件维度、页面维度，并提供统一的包来处理公用组件、特殊页面的通用逻辑，封装通用方法等，例如初始化浏览器信息、环境选择、登录、多网点切换、点击、输入、获取元素内容等等，业务回归用例只需要关注自己的用例操作步骤即可。 1、框架选择 – puppeteer[1]，它是由 Chrome 维护的 Node 库，基于 DevTools 协议来驱动 chrome 或者 chromium 浏览器运行，支持 headless 和 non-headless 两种方式。官网提供了非常丰富的文档，简单易学。 UI 自动化框架有很多种，包括 selenium、phantom；对比后发现 puppeteer 比较轻量，只需要增加一个 npm 包即可使用；它是基于事件驱动的方式，比 selenium 的等待轮询更稳当、性能更佳；另外，它是 chrome 原生支持，能提供所有 chrome 支持的 api，同时我们的业务场景只需要覆盖 chrome，所以它是最好的选择。 – mocha[2] + mochawesome[3]，mocha 是比较主流的测试框架，支持 beforeEach、before、afterEach、after 等钩子函数，assert 断言，测试套件，用例编排等。 mochawesome 是 mocha 测试框架的第三方插件，支持生成漂亮的 html/css 报告。 js 测试框架同样有很多可以选择，mocha、ava、Jtest 等等，选择 mocha 是因为它更灵活，很多配置可以结合第三方库，比如 report 就是结合了 mochawesome 来生成好看的 html 报告；断言可以用 powser-assert 替代。 2、脚本编写 封装基础库 封装 pc 端、h5 端浏览器的初始化过程 封装 pc 端、h5 端登录统一处理 封装页面模型和组件模型 封装上传组件、日期组件、select 组件等的统一操作方法 封装 input、click、hover、tap、scrollTo、hover、isElementShow、isElementExist、getElementVariable 等方法 提供根据 “html 标签>>页面文字” 形式获取页面元素及操作方法的统一支持 封装 baseTest，增加用例开始、结束后的统一操作 封装 assert，增加断言日志记录 业务用例 安装基础库 编排业务用例 3、执行逻辑 分环境执行 增加预上线环境代码变更触发、线上环境自动执行 监控源码变更 增加 gitlab webhook，监控开发源码合并 master 时自动在预上线环境执行 增加 gitlab webhook，监控测试用例变更时自动在生产环境执行 每日定时执行 增加 crontab，每日定时执行线上环境 [图片] [图片] [图片] [图片] 二、接口测试 接口测试主要针对于 Node 的 server 层，根据我们的开发规范，Node 不做复杂的业务逻辑，但是需要将服务化应用提供 dubbo 接口进行一次转换，或将多个 dubbo 接口组合起来，提供一个可供 h5/小程序渲染数据的 http 接口，转化过程就带来了各种数据的获取、组合、转换，形成了新的端到端接口。这个时候单单靠服务化接口的自动化已经不能保障对上层接口的全覆盖，所以我们针对 Node 接口也进行自动化测试。为了使用测试内部统一的测试框架，我们通过 java 去请求 Node 提供的 http 接口，那么当用例都写好之后，该如何评判接口测试的质量？是否完全覆盖了全部业务逻辑呢？此时就需要一个行之有效的方法来获取到测试的覆盖情况，以检查有哪些场景是接口测试中未覆盖的，做到更好的查漏补缺。 – istanbul[4] 是业界比较易用的 js 覆盖率工具，它利用模块加载的钩子计算语句、行、方法和分支覆盖率，以便在执行测试用例时透明的增加覆盖率。它支持所有类型的 js 覆盖率，包括单元测试、服务端功能测试以及浏览器测试。 但是，我们的接口用例写在 Java 代码中，通过 Http 请求的方式到达 Node 服务器，非 js 单测，也非浏览器功能测试，如何才能获取到 Node 接口的覆盖率呢？ 解决办法是增加 cover 参数：–handle-sigint，通过增加 --handle-sigint 参数启动服务，当服务接收到一个 SIGINT 信号（linux 中 SIGINT 关联了 Ctrl+C），会通知 istanbul 生成覆盖率。这个命令非常适合我们，并且因此形成了我们接口覆盖率的一个模型： [代码]1. istanbule --handle-sigint 启动服务 2. 执行测试用例 3. 发送 SIGINT结束istanbule，得到覆盖率 [代码] 最终，解决了我们的 Node 接口覆盖率问题，并通过 jenkins 持续集成来自动构建 [图片] [图片] [图片] 当然，在获取覆盖率的时候有需求文件是不需要统计的，可以通过在根路径下增加 .istanbule.yml 文件的方式，来排除或者指定需要统计覆盖率的文件 [代码]verbose: false instrumentation: root: . extensions: - .js default-excludes: true excludes:['**/common/**','**/app/constants/**','**/lib/**'] embed-source: false variable: __coverage__ compact: true preserve-comments: false complete-copy: false save-baseline: false baseline-file: ./coverage/coverage-baseline.json include-all-sources: false include-pid: false es-modules: false reporting: print: summary reports: - lcov dir: ./coverage watermarks: statements: [50, 80] lines: [50, 80] functions: [50, 80] branches: [50, 80] report-config: clover: {file: clover.xml} cobertura: {file: cobertura-coverage.xml} json: {file: coverage-final.json} json-summary: {file: coverage-summary.json} lcovonly: {file: lcov.info} teamcity: {file: null, blockName: Code Coverage Summary} text: {file: null, maxCols: 0} text-lcov: {file: lcov.info} text-summary: {file: null} hooks: hook-run-in-context: false post-require-hook: null handle-sigint: false check: global: statements: 0 lines: 0 branches: 0 functions: 0 excludes: [] each: statements: 0 lines: 0 branches: 0 functions: 0 excludes: [] [代码] 三、单元测试 单元测试在测试分层中处于金字塔最底层的位置，单元测试做的比较到位的情况下，能过滤掉大部分的问题，并且提早发现 bug，也可以降低 bug 成本。推行一段时间的单测后发现，在有赞的 Node 框架中，业务层的 server 端只做接口组装，client 端面向浏览器，都不太适合做单元测试，所以我们只针对基础框架和通用组件进行单测，保障基础服务可以通过单测排除大部分的问题。比如基础框架中店铺通用信息服务，单测检查店铺信息获取；比如页面级商品组件，单测检查商品组件渲染的 html 是否和原来一致。 单测方案试行了两个框架： Jest[5] ava[6] 比较推荐的是 Jest 方案，它支持 Matchers 方式断言；支持 Snapshot Testing，可测试组件类代码渲染的 html 是否正确；支持多种 mock，包括 mock 方法实现、mock 定时器、mock 依赖的 module 等；支持 istanbule，可以方便的获取覆盖率。 总之，前端的单测方案也越来越成熟，需要前端开发人员更加关注 js 单测，将 bug 扼杀在摇篮中。 四、基础库变更报警 上面我们已经对基础服务和基础组件进行了单元测试，但是单测也不能完全保证基础库的变更完全没有问题，伴随着业务层引入新版本的基础库，bug 会进一步带入到业务层，最终影响 C 端用户的正常使用。那如何保障每次业务层引入新版本的基础库之后能做到全面的回归？如何让业务测试同学对基础库变更更加敏感呢？针对这种情况，我们着手做了一个基础库版本变更的小工具。实现思路如下： [代码]1. 对比一次 master 代码的提交或 merge 请求，判断 package.json 中是否有特定基础库版本变更 2. 将对应基础库的前后两个版本的代码对比发送到测试负责人 3. 根据 changelog 判断此次回归的用例范围 4. 增加 gitlab webhook，只有合并到合并发布分支或者 master 分支的代码才触发检查 [代码] 这个小工具的引入能及时通知测试人员针对什么需求改动了基础组件，以及这次基础组件的升级主要影响了哪些方面，这样能避免相对黑盒的测试。 第一版实现了最简功能，后续再深挖需求，可以做到前端代码变更的精准测试。 [图片] 五、sentry 报警 在刚接触前端测试的时候，js 的报错没有任何追踪，对于排查问题和定位问题有很大困扰。因此我们着手引入了 sentry 报警监控，用于监控线上环境 js 的运行情况。 – sentry[7] 是一款开源的错误追踪工具，它可以帮助开发者实时监控和修复崩溃。 开始我们接入的方式比较简单粗暴，直接全局接入，带来的问题是报警信息非常庞大，全局上报后 info、warn 信息都会打出来。 更改后，使用 sentry 的姿势是： sentry 的全局信息上报，并进行筛选 错误类型: TypeError 或者 ReferenceError 错误出现用户 > 1k 错误出现在 js 文件中 出现错误的店铺 > 2家 增加核心业务异常流程的主动上报 最终将筛选后的错误信息通过邮件的形式发送给告警接收人，在固定的时间集中修复。 [图片] [图片] 六、业务报警 除了 sentry 监控报警，Node 接口层的业务报警同样是必不可少的一部分，它能及时发现 Node 提供的接口中存在的业务异常。这部分是开发和运维同学做的，包括在 Node 框架底层接入日志系统；在业务层正确的上报错误级别、错误内容、错误堆栈信息；在日志系统增加合理的告警策略，超过阈值之后短信、电话告警，以便于及时发现问题、排查问题。 业务告警是最能快速反应生产环境问题的一环，如果某次发布之后发生告警，我们第一时间选择回滚，以保证线上的稳定性。 七、约定规范 除了上述的一些测试和告警手段之外，我们也做了一些流程规范、用例维护等基础建设，包括： 发布规范 多个日常分支合并发布 限制发布时间 规范发布流程 整理自测核心检查要点 基线用例库 不同业务 P0 核心用例定期更新 项目用例定期更新到业务回归用例库 线上问题场景及时更新到回归用例库 目前有赞的前端测试套路基本就是这样，当然有些平时的努力没有完全展开，例如接口测试中增加返回值结构体对比；增加线上接口或页面的拨测[8]；给开发进行自测用例设计培训等等。也还有很多新功能探索中，如接入流量对比引擎，将线上流量导到预上线环境，在代码上线前进行对比测试；增加UI自动化的截图对比；探索小程序的UI自动化等等。 参考链接 [1] https://github.com/GoogleChrome/puppeteer [2] https://www.npmjs.com/package/mocha [3] https://www.npmjs.com/package/mochawesome [4] https://github.com/gotwarlost/istanbul [5] https://github.com/facebook/jest [6] https://github.com/avajs/ava [7] https://docs.sentry.io [8] https://tech.youzan.com/youzan-online-active-testing/