浏览器
01.实现全屏
当你需要将当前屏幕显示为全屏时。
function fullScreen() { const el = document.documentElement const rfs = el.requestFullScreen || el.webkitRequestFullScreen || el.mozRequestFullScreen || el.msRequestFullscreen if(typeof rfs != "undefined" && rfs) { rfs.call(el) }}// Click to go 🌐 full screen!fullScreen()02.退出全屏
当你需要退出全屏时。
function exitScreen() { if (document.exitFullscreen) { document.exitFullscreen() } else if (document.mozCancelFullScreen) { document.mozCancelFullScreen() } else if (document.webkitCancelFullScreen) { document.webkitCancelFullScreen() } else if (document.msExitFullscreen) { document.msExitFullscreen() } if(typeof cfs != "undefined" && cfs) { cfs.call(el) } // Exiting full screen 🚪🏃}exitScreen()03.页面打印
当你需要打印当前页面时。
// Click to print the page 🖨️window.print()
04.打印内容样式更改
当你需要打印出当前页面,但需要修改当前布局时。
<style>/* Adjusting print layout 🔧 */@media print { .noprint { display: none; }}</style><div class="print">print</div><div class="noprint">noprint</div>05.阻止关闭事件
当需要阻止用户刷新或关闭浏览器时,可以选择触发 beforeunload 事件,部分浏览器无法自定义文本内容。
window.onbeforeunload = function(){ return 'Are you sure you want to leave the haorooms blog?🚪🤔';};06.屏幕录制
当你需要录制当前屏幕并上传或下载屏幕录制时。
// Record your screen here const streamPromise = navigator.mediaDevices.getDisplayMedia()streamPromise.then(stream => { var recordedChunks = [];// recorded video datavar options = { mimeType: "video/webm; codecs=vp9" };// Set the encoding format var mediaRecorder = new MediaRecorder(stream, options);// Initialize the MediaRecorder instance mediaRecorder.ondataavailable = handleDataAvailable;// Set the callback when data is available (end of screen recording) mediaRecorder.start(); // Video Fragmentation function handleDataAvailable(event) { if (event.data.size > 0) { recordedChunks.push(event.data);// Add data, event.data is a BLOB object download();// Encapsulate into a BLOB object and download } }// file download function download() { var blob = new Blob(recordedChunks, { type: "video/webm" }); // Videos can be uploaded to the backend here var url = URL.createObjectURL(blob); var a = document.createElement("a"); document.body.appendChild(a); a.style = "display: none"; a.href = url; a.download = "test.webm"; a.click(); window.URL.revokeObjectURL(url); }})07.判断横竖屏
当你需要判断手机横竖屏的状态时。
function hengshuping() { if (window.orientation == 180 || window.orientation == 0) { alert("Portrait mode! 📱"); } if (window.orientation == 90 || window.orientation == -90) { alert("Landscape mode! 🌅"); }}// Add listener for screen orientation changes 🔄window.addEventListener("onorientationchange" in window ? "orientationchange" : "resize", hengshuping, false);08.改变横竖屏样式
当你需要为横竖屏设置不同的样式时。
<style>@media all and (orientation : landscape) { body { background-color: #ff0000; }}@media all and (orientation : portrait) { body { background-color: #00ff00; }}</style>09.标签页隐藏
当你需要监听标签页显示和隐藏事件时。
// Tab Page Hiddenconst {hidden, visibilityChange} = ( () => {let hidden, visibilityChange;if ( typeof document.hidden !== "undefined") {// Opera 12.10 and Firefox 18 and later supporthidden = "hidden";visibilityChange = "visibilitychange";} else if ( typeof document.msHidden !== "undefined") {hidden = "msHidden";visibilityChange = "msvisibilitychange";} else if ( typeof document.webkitHidden !== "undefined") {hidden = "webkitHidden";visibilityChange = "webkitvisibilitychange";}return {hidden,visibilityChange}})();const handleVisibilityChange = () => {console.log( "currently hidden", document[hidden]);};document.addEventListener(visibilityChange,handleVisibilityChange,false);
图片
10.本地图片预览
当你从客户端获取图片但无法立即上传到服务器,但需要预览时。
<!-- Upload and preview images here 📸 --><div class="test"> <input type="file" name="" id=""> <img src="" alt=""></div><script>const getObjectURL = (file) => { let url = null; if (window.createObjectURL != undefined) { // basic url = window.createObjectURL(file); } else if (window.URL != undefined) { // webkit or chrome url = window.URL.createObjectURL(file); } else if (window.URL != undefined) { // mozilla(firefox) url = window.URL.createObjectURL(file); } return url;}document.querySelector('input').addEventListener('change', (event) => { document.querySelector('img').src = getObjectURL(event.target.files[0])})</script>11.图片预加载
当你有大量图片时,你需要预加载图片以避免白屏
const images = []function preloader(args) { for (let i = 0, len = args.length; i < len; i++) { images[i] = new Image() images[i].src = args[i] } } preloader(['1.png', '2.jpg'])Js
12.字符串脚本
当需要将一串字符串转换成js脚本时,此方法存在xss漏洞,谨慎使用。
const obj = eval('({ name: "jack" })')// obj will be converted to object{ name: "jack" }const v = eval('obj')// v will become the variable obj13.递归函数名解耦
当需要编写递归函数时,会声明一个函数名,但每次修改函数名时,总会忘记修改内部的函数名。argument 是函数内部对象,包括传入函数的所有参数,arguments.callee 代表函数名。
// This is a basic Fibonacci sequencefunction fibonacci (n) { const fn = arguments.callee if (n <= 1) return 1 return fn(n - 1) + fn(n - 2)}DOM元素
14.隐式判断
当你需要判断某个dom元素当前是否出现在页面视图中时,可以尝试使用IntersectionObserver来判断。
< style>.item {height: 350px;}</ style>< div class= "container">< div class= "item" data-id= "1">Invisible </ div>< div class= "item" data-id= "2">Invisible </ div>< div class= "item" data-id= "3">Invisible </ div></ div>< script>if ( window?.IntersectionObserver) {let items = [...document.getElementsByClassName( "item")]; // parses as a true array, also available Array.prototype.slice.call()let io = new IntersectionObserver(( entries) => {entries.forEach( ( item) => {item.target.innerHTML =item.intersectionRatio === 1 // The display ratio of the element, when it is 1, it is completely visible, and when it is 0, it is completely invisible? `Element is fully visible`: `Element is partially invisible`;});},{root: null,rootMargin: "0px 0px",threshold: 1, // The threshold is set to 1, and the callback function is triggered only when the ratio reaches 1});items.forEach( ( item) => io.observe(item));}</ script>
15.元素可编辑
当你需要编辑 dom 元素时,让它像文本区域一样点击。
<div contenteditable="true">here can be edited</div>
16.元素属性监控
< div id= "test">test </ div>< button onclick= "handleClick()">OK </ button>< script>const el = document.getElementById( "test");let n = 1;const observe = new MutationObserver( ( mutations) => {console.log( "attribute is changede", mutations);})observe.observe(el, {attributes: true});function handleClick() {el.setAttribute( "style", "color: red");el.setAttribute( "data-name", n++);}setTimeout( () => {observe.disconnect(); // stop watch}, 5000);</ script>
17.打印dom元素
开发过程中需要打印dom元素时,使用console.log往往只能打印出整个dom元素,无法查看dom元素内部的属性,可以尝试使用console.dir。
console.dir(document.body)
其他
18.激活应用
当你在移动端开发时,需要打 开其他应用,以下方法也可以通过location.href赋值来操作
<!-- Use the following links to activate phone features 📲 --><a href="tel:12345678910">phone</a><a href="sms:12345678910,12345678911?body=hello">android message</a> <a href="sms:/open?addresses=12345678910,12345678911&body=hello">ios message</a><a href="wx://">ios message</a>
来自 “ ITPUB博客 ” ,链接:https://blog.itpub.net/70041921/viewspace-3040812/ ,如需转载,请注明出处,否则将追究法律责任。