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 crypto = require("node:crypto")
const AreaNameEnum = require('../enums/AreaNameEnum');
const Store = require("../utils/Store");
const JsUtil = require('../utils/JsUtil');
let store = new Store();
let sleep = ms => new Promise(resolve => setTimeout(resolve, ms));


router.post('/rsCookie', async (req, res) => {
    let uuid = crypto.randomUUID().replace(/-/g, "");
    let start = new Date();
    try {
        let url = req.body['url'];
        let areaName = req.body['areaName'];
        let htmlStr = req.body['htmlBase64'];
        let cookie = req.body['cookieBase64'];
        let userAgent = req.body['userAgentBase64'];
        console.log(`${uuid};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 cookies = await handle(url, uuid, areaName, Buffer.from(htmlStr, 'base64').toString('utf-8'), cookie != null && cookie !== "" ? Buffer.from(cookie, 'base64').toString('utf-8') : null, userAgent != null && userAgent !== "" ? Buffer.from(userAgent, 'base64').toString('utf-8') : null,)

        console.log(`${uuid};返回cookie ---->`, cookies)

        res.status(200).send(cookies);
    } catch (e) {
        console.error(e.stack)
        return res.status(500).send(e.message)
    } finally {
        console.log(`${uuid};rsCookie ${new Date() - start} ms`)
    }

})

function CookieStr2List(cookies) {
    let list = []
    for (let cookie of cookies.trim().split("; ")) {
        list.push(cookie);
    }
    return list
}

async function handle(url, uuid, areaName, htmlStr, cookie, userAgent) {
    // 获取 origin
    let baseUrl = new URL(url).origin;
    // 初始化 jsDom 和 cookieJar
    const [jsDom, cookieJar] = jsdomFromText({
        url: url, referrer: url, userAgent: userAgent, contentType: "text/html", runScripts: "outside-only", // runScripts: 'dangerously'/'outside-only'
    })
    // 加载dom
    let dom = await jsDom(htmlStr);
    console.log(`${uuid};html 加载长度--->`, dom.serialize().length)
    window = dom.window
    // ------------------------------------------------ param ----------------------------------------------------------
    // 标志判断cookie是否生成
    window[uuid] = false


    // ------------------------------------------------ function -------------------------------------------------------
    // js执行成功后会跳转页面 会触发onbeforeunload钩子
    window.onbeforeunload = async (url) => {
        console.debug(`${url} 页面回调完成`);
        window[uuid] = true
    }

    // 设置 cookie
    if (cookie != null) {
        let cookieList = CookieStr2List(cookie);
        console.log(`${uuid};cookie 加载长度--->`, cookieList.length, baseUrl)
        for (let i = 0; i < cookieList.length; i++) {
            cookieJar.setCookieSync(cookieList[i], baseUrl);
        }
    }

    // 方案1 通过监听cookie 判断cookie是否生成
    const superSetCookie = cookieJar.setCookie;
    // 设置 setCookie 代理
    cookieJar.setCookie = function (cookie, currentUrl, options, callback) {
        console.debug(`${uuid};正在设置 Cookie:`, cookie, currentUrl);
        let call = superSetCookie.call(this, cookie, currentUrl, options, callback);
        // 设置标志可取标志
        if (cookie.includes('YqQ7a3SgknV8P')) {
            window[uuid] = true
        }
        return call;
    };

    // ------------------------------------------------ 实例化浏览器 -----------------------------------------------------
    browser(window, 'chrome');
    // 加载js
    let js = await JsUtil.loadJs(window.document, areaName, cookie);
    console.log(`${uuid};js 加载长度--->`, js.length)


    // 执行 js
    let script = new Script(js);
    let internalVMContext = dom.getInternalVMContext();
    script.runInContext(internalVMContext, {timeout: 1000});

    // 等待cookie 被设置
    for (let i = 0; i < 10; i++) {
        if (window[uuid]) {
            break;
        }
        await sleep(100)
    }
    // 获取cookie
    let resCookie = cookieJar.getCookieStringSync(baseUrl);
    // 关闭
    dom.window.close()
    return resCookie;
}

/**
 * 方案1
 * 利用rs特性 js加载完成后会刷新页面
 * 利用 window.onbeforeunload 判断页面是否执行完毕
 * @param window
 * @param cookieJar
 * @param uuid
 * @returns {Promise<*|null>}
 */
function scheme1_before(window, cookieJar, uuid) {
    window.onbeforeunload = async (url) => {
        let baseUrl = new URL(url).origin;
        const cookies = cookieJar.getCookieStringSync(baseUrl);
        console.debug(`${url} 页面回调生成cookie：`, cookies);
        store.set(uuid, cookies)
        // window.close();
    }
}

/**
 * 方案1
 * @param window
 * @param cookieJar
 * @param uuid
 * @returns {Promise<*|null>}
 */
async function scheme1_after(window, cookieJar, uuid) {
    // 等待 onbeforeunload 钩子触发后的回掉
    let val = await store.waitGetAndDelete(uuid, 100, 10)
    if (val != null) {
        return val;
    }
    return null;
}


/**
 * 监听cookie出现指定cookie
 * @param window
 * @param cookieJar
 * @param uuid
 * @returns {Promise<null|string>}
 */
function scheme2_before(window, cookieJar, uuid) {
    // window[uuid] = false;
    const superSetCookie = cookieJar.setCookie;
    // 设置 setCookie 代理
    cookieJar.setCookie = function (cookie, currentUrl, options, callback) {
        console.log(`${uuid};正在设置 Cookie:`, cookie, currentUrl);
        return superSetCookie.call(this, cookie, currentUrl, options, callback);
        // let cookieStringSync = super.getCookieStringSync();
        // console.log(cookieStringSync)
        // if (cookie.includes(key)) {
        //     // 设置标志可取标志
        //     window[uuid] = true
        // }
        // return call;
    };
}

/**
 * 方案2
 * @param window
 * @param cookieJar
 * @param key
 * @param uuid
 * @returns {Promise<null|string>}
 */
async function scheme2_after(window, cookieJar, key, uuid) {
    let sleep = ms => new Promise(resolve => setTimeout(resolve, ms));
    for (let i = 0; i < 10; i++) {
        let cookieStringSync = cookieJar.getCookieStringSync();
        if (cookieStringSync.includes(key)) {
            return cookieStringSync;
        }
        await sleep(100)
    }
    return null;
}

/**
 * 方案3 根据cookie现有的数量判断
 * @param window
 * @param cookieJar
 * @param baseUrl
 * @param uuid
 */
function scheme3_before(window, cookieJar, baseUrl, uuid) {
    const initCookie = cookieJar.getCookieStringSync(baseUrl);
    window[uuid + 'CookieSize'] = initCookie != null ? initCookie.trim().split("; ").length : 0;
}

/**
 * 方案3
 * @param window
 * @param cookieJar
 * @param baseUrl
 * @param uuid
 * @returns {Promise<null|string>}
 */
async function scheme3_after(window, cookieJar, baseUrl, uuid) {
    let initCookieLength = window[uuid + 'CookieSize']
    let sleep = ms => new Promise(resolve => setTimeout(resolve, ms));
    for (let i = 0; i < 10; i++) {
        let cookieStringSync = cookieJar.getCookieStringSync(baseUrl);
        let cookies = cookieStringSync != null ? cookieStringSync.trim().split("; ").length : 0;
        if (cookies > initCookieLength) {
            return cookieStringSync;
        }
        await sleep(100)
    }
    return null;
}


module.exports = router