一种在线词典简易爬虫

Cambridge Dictionary

爬虫规则：

1. 优先选择 English-Chinese 页面，若没有，则选择 English 页面。
2. 去除了多余的下划线、“Add to word list”按钮、播放按钮等元素。

使用方法：

1. 打开 Cambridge Dictionary。
2. 打开浏览器的控制台。方法 1：页面空白位置右键，检查元素；方法 2：按 F12。
3. 选择 Console 一栏，粘贴下面的代码并运行。
4. 在新出现的最右边的框中输入需要整理的英文单词，一行一个。
5. 点击“work”按钮，爬虫开始采集。
6. 爬虫采集完毕（出现“OK!”字样）时，点击“print”按钮，开始打印页面。

function ready(){
	document.body.innerHTML="<iframe id=\"iframe\" src=\"\" height=\"500px\" width=\"500px\"></iframe><textarea id=\"output\" style=\"height:500px;\" readonly></textarea><textarea id=\"error\" style=\"height:500px;\" readonly></textarea><textarea id=\"input\" style=\"height:500px;\" placeholder=\"input your word list here\"></textarea><button onclick=\"work();\">work</button><button onclick=\"print();\">print</button>";
}
ready();
function delClassName(iframe,name){let a=iframe.getElementsByClassName(name);while(a.length)a[0].parentNode.removeChild(a[0]);}
function eraseUnderlineClassName(iframe,name){let a=iframe.getElementsByClassName(name);for(let i=0;i<a.length;i++){let b=a[i];while(b!=iframe.body){b.style.textDecoration="none";b=b.parentNode;}}}
function copy(val){
	var node_textarea=document.createElement("textarea");
	node_textarea.style.opacity=0;
	node_textarea.value=val;
	node_textarea.readonly=true;
	document.body.appendChild(node_textarea);
	node_textarea.select();
	document.execCommand("Copy");
	document.body.removeChild(node_textarea);
}
var words,id;
function next_word(){
	id++;
	loadIframe();
}
function extract(){
	if(document.getElementById("iframe").src.slice(0,4)!="http")return;
	let iframe=document.getElementById("iframe").contentWindow.document;
	iframe.body.innerHTML=iframe.getElementsByClassName("entry")[0].outerHTML;
	delClassName(iframe,"daud");
	delClassName(iframe,"daccord");
	delClassName(iframe,"dwl");
	delClassName(iframe,"def-info");
	delClassName(iframe,"lb-cm");
	delClassName(iframe,"dphrase-info");
	delClassName(iframe,"phrasal_verb");
	delClassName(iframe,"phrasal_verbs");
	delClassName(iframe,"idiom");
	delClassName(iframe,"idioms");
	delClassName(iframe,"related_word");
	delClassName(iframe,"related_words");
	delClassName(iframe,"dbtn");
	eraseUnderlineClassName(iframe,"query");
	eraseUnderlineClassName(iframe,"dtrans");
	//copy(iframe.body.innerHTML);
	document.getElementById("output").value+=iframe.body.innerHTML;
	next_word();
}
function loadIframe(){
	if(id==words.length||(id==words.length-1&&words[id]=="")){
		document.getElementById("error").value+="OK!\n";
		return;
	}
	let word=words[id];
	document.getElementById("iframe").src="https://dictionary.cambridge.org/dictionary/english-chinese-simplified/"+word;
	setTimeout(function(){
		let iframe=document.getElementById("iframe").contentWindow.document;
		if(iframe.baseURI=="https://dictionary.cambridge.org/dictionary/english-chinese-simplified/"){
			document.getElementById("error").value+="Failed to find English-Chinese interpretation of the word "+word+".\n";
			document.getElementById("iframe").src="https://dictionary.cambridge.org/dictionary/english/"+word;
			setTimeout(function(){
				let iframe=document.getElementById("iframe").contentWindow.document;
				console.log(iframe.baseURI);
				if(iframe.baseURI=="https://dictionary.cambridge.org/dictionary/english/"){
					document.getElementById("error").value+="Failed to find the word "+word+".\n";
					next_word();
				}else extract();
			},3000);
		}else extract();
	},3000);
}
function work(){
	words=document.getElementById("input").value.split("\n");
	id=0;
	console.log(words);
	document.getElementById("output").value="";
	document.getElementById("error").value="";
	loadIframe();
}
function print(){
	let iframe=document.getElementById("iframe").contentWindow;
	iframe.document.body.innerHTML=document.getElementById("output").value;
	iframe.print();
}