Node爬虫项目精简版 wallhaven网站实操 2023.8.29

练习地址： https://wallhaven.cc/toplist

const express = require('express');
const axios = require('axios');
const cheerio = require('cheerio');
const schedule = require('node-schedule');
const fs = require('fs');

async function downloadImage(url) {
    const response = await axios.get(url, {responseType: 'arraybuffer'});
    return response.data;
}
async function saveImage(url, path) {
    const image = await downloadImage(url);
    fs.writeFileSync(path, image);
}
async function AccessToResources(url) {
    try {
        const response = await axios.get(url);
        const $ = cheerio.load(response.data);
        const images = [];
        $('a.preview').each((i, el) => {
            const img = $(el).attr('href');
            if (!img) return;
            images.push(img);
        });
        for (let i = 0; i < images.length; i++) {
            const date = new Date();
            const response = await axios.get(images[i]);
            const $ = cheerio.load(response.data);
            const img = $('img#wallpaper').attr('src');
            const fileName = date.getTime();
            await saveImage(img, `./images/${fileName}.jpg`)
        }
    } catch (e) {
        console.log('e');
    }
}

const rule = new schedule.RecurrenceRule();
rule.hour = [1, 2, 3, 4, 5, 6];
rule.minute = 0;
const url = 'https://wallhaven.cc/toplist';
const job = schedule.scheduleJob(rule, function () {
    console.log('开始执行任务', rule.hour)
    AccessToResources(url);
});
AccessToResources(url);

const app = express();
app.use(express.static('images'));
app.get('/', (req, res) => {
    res.send('Hello World!');
})
app.listen(1234, () => {
    console.log('server is running at http://localhost:1234');
})