spider_creater/backend/vm/load_content.js

var ret = {}
var tmp = null

if ("{{.TitleCss}}" != "") {//标题
	tmp = document.querySelector("{{.TitleCss}}")
	if (tmp) ret["title"] = tmp.getAttribute("title") || tmp.innerText
}
if ("{{.PublishUnitCss}}" != "") {//采购单位
	tmp = document.querySelector("{{.PublishUnitCss}}")
	if (tmp) ret["publishUnit"] = tmp.getAttribute("title") || tmp.innerText
}
if ("{{.PublishTimeCss}}" != "") {//发布时间
	tmp = document.querySelector("{{.PublishTimeCss}}")
	if (tmp) ret["publishTime"] = tmp.getAttribute("title") || tmp.innerText
}
if ("{{.ContentCss}}" != "") {//正文内容
	tmp = document.querySelector("{{.ContentCss}}")
	if (tmp) {
		ret["content"] = tmp.innerText
		ret["contentHtml"] = tmp.innerHTML
		var patchContent = false
		//处理详情页中的大图，大图作为附件使用
		const images = tmp.querySelectorAll("img");
		images.forEach((img, i) => {
			if (img.width > 300) {
				patchContent = true
				const a = document.createElement("a");
				a.href = img.src;
				a.innerText = img.src;
				tmp.appendChild(a);
			}
		})
	}
}
if("{{.AttachCss}}"!=""){//附件
	tmp = document.querySelectorAll("{{.AttachCss}} a")
	let attach=[]
	if(tmp){
		tmp.forEach((v,i)=>{
			attach.push({title:v.getAttribute("title")||v.innerText,href:v.href})
		})
	}
	ret["attachLinks"]=attach
}
//检查中文字符个数，少于20，修正正文内容
let regex = /[\u4e00-\u9fa5]/g;
let chineseCharacters = ret["content"]?ret["content"].match(regex):[];
let chineseCharactersLen=chineseCharacters ? chineseCharacters.length : 0;
if (chineseCharactersLen < 20 && ret["attachLinks"] && ret["attachLinks"].length>0) ret["content"] = '详情请访问原网页！'
ret