qry-python-js/routes/index.js

const express = require("express");
const router = express.Router();

const {jsdomFromText, browser} = require("sdenv");
const {Script} = require("node:vm");
const fs = require("node:fs");
const crypto = require("node:crypto")
const AreaNameEnum = require('../enums/AreaNameEnum');
const Store = require("../utils/Store");
let store = new Store();


router.post('/rsCookie', async (req, res) => {
    let start = new Date();
    try {
        let url = req.body['url'];
        let areaName = req.body['areaName'];
        let htmlStr = req.body['htmlBase64'];
        let jsStr = req.body['jsBase64'];
        let cookie = req.body['cookieBase64'];
        let userAgent = req.body['userAgentBase64'];
        console.log(`接收到 ${areaName} 请求：${url}`)
        if (url == null || url === '') {
            return res.status(500).send('error url')
        }
        if (htmlStr == null || htmlStr === '') {
            return res.status(500).send('error html')
        }

        let jsText;
        if (jsStr == null || jsStr === "") {
            let jsPath = AreaNameEnum.getByAreaName(areaName).JS_FILE
            if (jsPath == null) {
                console.error('未找到js文件')
                return res.send('未找到js文件')
            }
            jsText = fs.readFileSync(jsPath).toString('utf8');
        } else {
            jsText = Buffer.from(jsStr, 'base64').toString('utf-8')
        }


        let cookies = await handle(url,
            Buffer.from(htmlStr, 'base64').toString('utf-8'),
            jsText,
            cookie != null && cookie !== "" ? Buffer.from(cookie, 'base64').toString('utf-8') : null,
            userAgent != null && userAgent !== "" ? Buffer.from(userAgent, 'base64').toString('utf-8') : null)
        res.status(200).send(cookies);
    } catch (e) {
        console.error(e)
        return res.status(500).send(e.toString())
    } finally {
        console.log(`rsCookie ${new Date() - start} ms`)
    }

})

function loadJs(window, jsText) {
    // 加载js
    let js = '';
    // 加载 页面上的js
    const allScript = window.document.querySelectorAll('script[r="m"]');
    for (let i = 0; i < allScript.length; i++) {
        const script = allScript[i];
        let attr = script.textContent;
        if (attr) {
            js += attr
        } else {
            js += jsText
        }
        js += ";\n"
    }
    return js;
}

async function handle(url, htmlStr, jsText, cookie, userAgent) {
    let uuid = crypto.randomUUID()
    // 获取 origin
    let baseUrl = new URL(url).origin;
    // 初始化 jsDom 和 cookieJar
    const [jsDom, cookieJar] = await jsdomFromText({
        url: url,
        referrer: url,
        userAgent: userAgent,
        contentType: "text/html",
        runScripts: "outside-only", // runScripts: 'dangerously'/'outside-only'
    })
    // 设置 cookie
    if (cookie != null) {
        console.log('cookie 加载长度--->', cookie, baseUrl)
        cookieJar.setCookieSync(cookie, baseUrl);
    }
    // 加载dom
    let dom = await jsDom(htmlStr);
    // console.log('html 加载长度--->', dom.serialize())
    console.log('html 加载长度--->', dom.serialize().length)

    window = dom.window
    // js执行成功后的瑞树会跳转页面 会触发onbeforeunload钩子
    window.onbeforeunload = async (url) => {
        const cookies = cookieJar.getCookieStringSync(baseUrl);
        // console.debug(`${url} 生成cookie：`, cookies);
        store.set(uuid, cookies)
        // window.close();
    }
    // 初始化浏览器
    browser(window, 'chrome');
    // 加载js
    let js = loadJs(window, jsText);
    console.log('js 加载长度--->', js.length)

    // 执行 js
    let script = new Script(js);
    let internalVMContext = dom.getInternalVMContext();
    script.runInContext(internalVMContext);
    // 等待 onbeforeunload 钩子触发后的回掉
    let timeout = -1;
    let reTry = -1;
    if (process.env.NODE_ENV === 'prod') {
        timeout = 100
        reTry = 10
    }
    let val = await store.waitGetAndDelete(uuid, timeout, reTry)

    internalVMContext.close()
    window.close()
    dom = null

    if (val != null) {
        return val;
    }
    throw new Error('执行超时')
}


module.exports = router