【Node.js】爬虫

安装：npm i puppeteer

就是自动控制浏览器进行点击，获取元素等操作。

import puppeteer  from "puppeteer";
// puppeteer 每个操作都是异步的

const input = process.argv[2]
// 默认无头模式，浏览器在后台操作
const browser = await puppeteer.launch({
    headless: false,
    executablePath: 'C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe'  // 使用浏览器的路径(默认为Chrome)
})
// 开启页面
const page = await browser.newPage()
// 进行跳转
await page.setViewport({width: 1920, height: 1000})
await page.goto('https://www.zhihu.com/')
// 获取元素
await page.waitForSelector('.css-16vu25q')
const elements = await page.$$('.css-16vu25q .GlobalWriteV2-navTop .GlobalWriteV2-topTitle')
const collectContent = async () => {
    const titleList = []
    await page.waitForSelector('. css-0')
    const elements = await page.$$('. css-0 .css-1rs9lm3 .css-1o7uqnr')
    for await (const el of elements) {
        const prop = await el.getProperty('innerText')
        const text = await prop.jsonValue()
        titleList.push(text)
    }
    console.log(titleList)
}
for await (const el of elements) {
    const prop = await el.getProperty('innerText')
    const text = await prop.jsonValue()
    console.log(text)
    if (text === (input || '回答问题')) {
        await el.click()
        collectContent()
        break
    }
}