修改爬虫

This commit is contained in:
编码猿
2024-11-22 12:31:44 +08:00
parent 07c46ac300
commit 8fdd65b9bb
11 changed files with 2551 additions and 2 deletions

43
newBackApi/router/info.js Normal file
View File

@@ -0,0 +1,43 @@
import express from "express";
// import app from "./utils/app.js"
import { Crawl } from '../utils/crawl.js'
import Config from '../config/config.js'
const router = express.Router();
const crawlApp = new Crawl()
// http://192.168.31.101:3000/api/v1/info/list?id=6808
router.get('/list', async (req, res) => {
let { id } = req.query
const { browser, page } = await crawlApp.page(`${Config.baseUrl}/${id}.html`);
await page.waitForSelector('.article')
let result = {
title: await page.$eval('.article-header h1 a', el => el.innerText),
src_list: await page.$$eval('.article-content .theme-box p img', el => el.map(x => { return { title: x.getAttribute("alt"), alt: x.getAttribute("data-src")} })),
recommend: []
}
const recommendList = await page.$$('.zib-widget .swiper-wrapper .swiper-slide')
for (const handle of recommendList) {
result.recommend.push({
id: await handle.$eval('a', el => el.pathname.replace(/[^\d]/g, "")),
thumbnail: await handle.$eval('a img', el => el.getAttribute("data-src")),
title: await handle.$eval('a img+div', el => el.innerText),
time: await handle.$eval('a .px12 item', el => el.innerText),
view: await handle.$eval('a .px12 .pull-right', el => el.innerText),
})
}
res.json({
code: 300,
data: result
})
})
export default router