---

效果展示

说明

• 看了看网上很少做这个功能，但是我有这个需求，就抽出事件写了个简单的工具
• 目前只能导出专栏的文章，不过思路类似
• 并没有做 Promise 失败的重新发送，代码仍然待完善，欢迎提交分支 PR，仓库地址：github 代码仓库地址

利用工具

• puppeteer：自动化库
• asyncPool：并发控制库

整体思路

1. 获取到你想提取的文章 id
2. 根据 id 打开编辑器，利用自带的导出按钮

代码

// index.js
import puppeteer from "puppeteer";
import asyncPool from "tiny-async-pool";
import { getPage, waitingOpenURL, findElement, clickImport } from "./tools.js";(async () => {// 关闭无头模式，显示浏览器窗口// userDataDir 表示把登录信息放到当前目录下，省着我们每次调用脚本都需要登录const browser = await puppeteer.launch({headless: false,userDataDir: "./userData",});const page = await browser.newPage();page.on("dialog", async (dialog) => {await dialog.accept();});let targetURL = "https://blog.csdn.net/u010263423/category_9162796.html";await page.goto(targetURL);await page.setViewport({ width: 1080, height: 1024 });const targetPageCount = await getPage(page);const willOpenArr = await waitingOpenURL(targetPageCount, targetURL);const findArray = [];findArray.push(...(await findElement(page)));if (targetPageCount > 1) {for (let i = 0; i < willOpenArr.length; i++) {await page.goto(willOpenArr[i]);findArray.push(...(await findElement(page)));}}const baseWriteURL = `https://editor.csdn.net/md/?articleId=`;const baseWriteURLArray = findArray.map((i) => `${baseWriteURL}${i.id}`);let successHandle = 0;function handleURL(url) {return new Promise(async (resolve) => {const page = await browser.newPage();page.on("dialog", async (dialog) => {await dialog.accept();});await page.goto(url);await clickImport(page);await page.close();await new Promise((r) => setTimeout(r, 300));resolve(`${url} 解析完成 ${++successHandle}`);});}for await (const ms of asyncPool(2, baseWriteURLArray, handleURL)) {console.log(ms);}console.log("***已完成所有解析***");
})();

// tools.js
/*** 功能：获取文章标题* @param {*} lis* @returns*/
export async function getTitle(lis) {const titles = await lis.$$eval(".title", (elements) => {return elements.map((e) =>e.innerHTML.replace("\n", "").split("<!--####试读-->")[0].replace("\n", "").trim());});return titles;
}/*** 功能：获取文章写作时间* - nth-child 选择器从1开始，前面尽量是标签名吧，如果是类的话，我试了一下选择不到* @param {*} lis*/
export async function getDate(lis) {const titleDate = await lis.$$eval(".column_article_data span:nth-child(2)",(elements) => {return elements.map((e) => e.innerHTML.trim().split(" &nbsp")[0]);});return titleDate;
}export async function getID(lis) {const titleId = await lis.$$eval("a", (elements) => {return elements.map((e) => e.href.split("details/")[1]);});return titleId;
}export async function getPage(page) {const pageContainer = await page.$(".ui-paging-container");let pageCount = 1;if (pageContainer) {const pageContext = await pageContainer.$$eval(".ui-pager", (elements) => {return elements.map((e) => e.innerHTML);});pageCount = Number(pageContext[pageContext.length - 3]);}console.log(pageCount);return pageCount;
}export async function waitingOpenURL(targetPageCount, targetURL) {const arr = [];if (targetPageCount > 1) {for (let i = 2; i <= targetPageCount; i++) {const front = targetURL.split(".html")[0];const url = `${front}_${i}.html`;arr.push(url);}}return arr;
}export async function findElement(page) {// 等待页面选择器的出现await page.waitForSelector(".column_article_list");const lis = await page.$(".column_article_list");// 获取文章标题、写作时间、文章idconst titles = await getTitle(lis);const titleId = await getID(lis);const titleDate = await getDate(lis);// 整理成数组对象const notes = [];titles.forEach((item, index) => {const obj = {title: item,date: titleDate[index],id: titleId[index],};notes.push(obj);});return notes;
}/*** 功能: 点击导出按钮* @param {*} page*/
export async function clickImport(page) {await new Promise((r) => setTimeout(r, 1500));const importButton ="div.layout__panel.layout__panel--navigation-bar.clearfix > nav > div.scroll-box > div:nth-child(1) > div:nth-child(22) > button";await page.waitForSelector(importButton);await page.click(importButton);// await new Promise((r) => setTimeout(r, 500));const nextImportButton ="div.side-bar__inner > div.side-bar__panel.side-bar__panel--menu > a:nth-child(1)";await page.waitForSelector(nextImportButton);await page.click(nextImportButton);// 这个时间是不能省的，一定要给点击事件留点时间，// 不然直接跳转页面，下载就失效了await new Promise((r) => setTimeout(r, 500));
}export function handleWriteURLs(url) {return new Promise((resolve) => {console.log(url);resolve();});
}