web逆向-猿人学第一题-js混淆、源码乱码

猿人学第一关：js 混淆 - 源码乱码

之前第一次做这道题的时候，卡了一晚上，现在重新来分析看看，首先是题目：

先打开开发者工具，看看数据来源，这里有作者设置了 setInterval 的反调试，每隔一段时间就会自动下断点，我们直接在 debugger 那一行 never pause here

然后点击 resume，以后就会自动跳过这一行了，来看一下数据接口

可以看到数据是以json的形式直接返回的，但这里有个 m参数，不确定是怎么来的

1	m: c204206e7e8953fbba7c33835c5478a9丨1654198904

长度是32，看上去好像是 md5 | 时间戳 的形式，下断点分析一波：

追栈到这里就差不多了，但源码稍微经过了混淆，不太容易看，还原出部分比较重要的源码：

window["url"] = "/api/" + "match" + "/1", request = function () {
    var _0x2268f9 = Date["parse"](new Date()) + (16798545 + -72936737 + 156138192), _0x57feae = oo0O0(_0x2268f9["toStr" + "ing"]()) + window["f"];
    const _0x5d83a3 = {};
    _0x5d83a3["page"] = window["page"], _0x5d83a3["m"] = _0x57feae + "丨" + _0x2268f9 / (-1 * 3483 + -9059 + 13542);
    var _0xb89747 = _0x5d83a3;

注意这一句 _0x5d83a3["m"] = _0x57feae + "丨" + _0x2268f9 / (-1 * 3483 + -9059 + 13542) ，其中：

(-1 * 3483 + -9059 + 13542) 的值是 1000

_0x2268f9 则是

1	Date["parse"](new Date()) + (16798545 + -72936737 + 156138192)

其实就是当前的 毫秒时间戳 + 100000000，也就是

1	int(time.time()) * 1000 + 100000000

到这里就剩 _0x57feae 还没分析了，上面的源码里面有：

1	_0x57feae = oo0O0(_0x2268f9["toStr" + "ing"]()) + window["f"]

控制台打印一下，window["f"] 里面就有我们想要的值，那前面的 oo0O0 函数难道是多余的，用来掩人耳目？

在关键代码处打断点重新调试一下：

所以这里基本可以确定，执行完 oo0O0 函数后，window["f"] 的值发生了改变，进到函数内部看看：

代码被设置了防格式化，我们直接抠出来，用工具美化一下

function oo0O0(mw) {
    window.b = '';
    for (var i = 0, len = window.a.length; i < len; i++) {
        console.log(window.a[i]);
        window.b += String[document.e + document.g](window.a[i][document.f + document.h]() - i - window.c)
    }
    var U = ['W5r5W6VdIHZcT8kU', 'WQ8CWRaxWQirAW=='];
    var J = function (o, E) {
        o = o - 0x0;
        var N = U[o];
        if (J['bSSGte'] === undefined) {
            var Y = function (w) {
                var m = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789+/=',
                    T = String(w)['replace'](/=+$/, '');
                var A = '';
                for (var C = 0x0, b, W, l = 0x0; W = T['charAt'](l++); ~W && (b = C % 0x4 ? b * 0x40 + W : W, C++
                        % 0x4) ? A += String['fromCharCode'](0xff & b >> (-0x2 * C & 0x6)) : 0x0) {
                    W = m['indexOf'](W)
                }
                return A
            };
            var t = function (w, m) {
                var T = [],
                    A = 0x0,
                    C, b = '',
                    W = '';
                w = Y(w);
                for (var R = 0x0, v = w['length']; R < v; R++) {
                    W += '%' + ('00' + w['charCodeAt'](R)['toString'](0x10))['slice'](-0x2)
                }
                w = decodeURIComponent(W);
                var l;
                for (l = 0x0; l < 0x100; l++) {
                    T[l] = l
                }
                for (l = 0x0; l < 0x100; l++) {
                    A = (A + T[l] + m['charCodeAt'](l % m['length'])) % 0x100, C = T[l], T[l] = T[A], T[A] = C
                }
                l = 0x0, A = 0x0;
                for (var L = 0x0; L < w['length']; L++) {
                    l = (l + 0x1) % 0x100, A = (A + T[l]) % 0x100, C = T[l], T[l] = T[A], T[A] = C, b += String[
                        'fromCharCode'](w['charCodeAt'](L) ^ T[(T[l] + T[A]) % 0x100])
                }
                return b
            };
            J['luAabU'] = t, J['qlVPZg'] = {}, J['bSSGte'] = !![]
        }
        var H = J['qlVPZg'][o];
        return H === undefined ? (J['TUDBIJ'] === undefined && (J['TUDBIJ'] = !![]), N = J['luAabU'](N, E), J[
            'qlVPZg'][o] = N) : N = H, N
    };
    eval(atob(window['b'])[J('0x0', ']dQW')](J('0x1', 'GTu!'), '\x27' + mw + '\x27'));
    return ''
}

把上面那段代码拿到 snippet 调试一下

注意需要和网站源码保持在同一页，不然有些参数是 undefined，比如 window.a 等等

调试到最后一行，这个地方有点可疑：

atob(window['b'])还原出来居然是一段代码，简直 amazing！原理的话其实就是把经过 base64编码的一整段代码放在 window['b'] 里面，再来看其他几个参数的还原：

这里是做一个字符串的替换，其实就等同于：

1	.replace('mwqqppz', "'" + mw + "'")

这里 mw的值是 1654200848000，也就是把 window.f = hex_md5(mwqqppz) 替换成了 window.f = hex_md5('1654200848000')

进入到 window[‘b’] 内部看一下，找到最后一行代码：

已经可以确定 window.f 的生成点了，来验证一下 hex_md5的真实性

对比发现，用 python标准库计算的md5值和 hex_md5的还不太一样，所以这里的md5应该是经过了作者的魔改

那没办法用python还原了，只能把 hex_md5 整个抠出来，用 node 子进程来执行，得到结果

var hexcase = 0;
var b64pad = "";
var chrsz = 16;

function hex_md5(a) {
    return binl2hex(core_md5(str2binl(a), a.length * chrsz))
}

function b64_md5(a) {
    return binl2b64(core_md5(str2binl(a), a.length * chrsz))
}

function str_md5(a) {
    return binl2str(core_md5(str2binl(a), a.length * chrsz))
}

function hex_hmac_md5(a, b) {
    return binl2hex(core_hmac_md5(a, b))
}

function b64_hmac_md5(a, b) {
    return binl2b64(core_hmac_md5(a, b))
}

function str_hmac_md5(a, b) {
    return binl2str(core_hmac_md5(a, b))
}

function md5_vm_test() {
    return hex_md5("abc") == "900150983cd24fb0d6963f7d28e17f72"
}

function core_md5(p, k) {
    p[k >> 5] |= 128 << ((k) % 32);
    p[(((k + 64) >>> 9) << 4) + 14] = k;
    var o = 1732584193;
    var n = -271733879;
    var m = -1732584194;
    var l = 271733878;
    for (var g = 0; g < p.length; g += 16) {
        var j = o;
        var h = n;
        var f = m;
        var e = l;
        o = md5_ff(o, n, m, l, p[g + 0], 7, -680976936);
        l = md5_ff(l, o, n, m, p[g + 1], 12, -389564586);
        m = md5_ff(m, l, o, n, p[g + 2], 17, 606105819);
        n = md5_ff(n, m, l, o, p[g + 3], 22, -1044525330);
        o = md5_ff(o, n, m, l, p[g + 4], 7, -176418897);
        l = md5_ff(l, o, n, m, p[g + 5], 12, 1200080426);
        m = md5_ff(m, l, o, n, p[g + 6], 17, -1473231341);
        n = md5_ff(n, m, l, o, p[g + 7], 22, -45705983);
        o = md5_ff(o, n, m, l, p[g + 8], 7, 1770035416);
        l = md5_ff(l, o, n, m, p[g + 9], 12, -1958414417);
        m = md5_ff(m, l, o, n, p[g + 10], 17, -42063);
        n = md5_ff(n, m, l, o, p[g + 11], 22, -1990404162);
        o = md5_ff(o, n, m, l, p[g + 12], 7, 1804660682);
        l = md5_ff(l, o, n, m, p[g + 13], 12, -40341101);
        m = md5_ff(m, l, o, n, p[g + 14], 17, -1502002290);
        n = md5_ff(n, m, l, o, p[g + 15], 22, 1236535329);
        o = md5_gg(o, n, m, l, p[g + 1], 5, -165796510);
        l = md5_gg(l, o, n, m, p[g + 6], 9, -1069501632);
        m = md5_gg(m, l, o, n, p[g + 11], 14, 643717713);
        n = md5_gg(n, m, l, o, p[g + 0], 20, -373897302);
        o = md5_gg(o, n, m, l, p[g + 5], 5, -701558691);
        l = md5_gg(l, o, n, m, p[g + 10], 9, 38016083);
        m = md5_gg(m, l, o, n, p[g + 15], 14, -660478335);
        n = md5_gg(n, m, l, o, p[g + 4], 20, -405537848);
        o = md5_gg(o, n, m, l, p[g + 9], 5, 568446438);
        l = md5_gg(l, o, n, m, p[g + 14], 9, -1019803690);
        m = md5_gg(m, l, o, n, p[g + 3], 14, -187363961);
        n = md5_gg(n, m, l, o, p[g + 8], 20, 1163531501);
        o = md5_gg(o, n, m, l, p[g + 13], 5, -1444681467);
        l = md5_gg(l, o, n, m, p[g + 2], 9, -51403784);
        m = md5_gg(m, l, o, n, p[g + 7], 14, 1735328473);
        n = md5_gg(n, m, l, o, p[g + 12], 20, -1921207734);
        o = md5_hh(o, n, m, l, p[g + 5], 4, -378558);
        l = md5_hh(l, o, n, m, p[g + 8], 11, -2022574463);
        m = md5_hh(m, l, o, n, p[g + 11], 16, 1839030562);
        n = md5_hh(n, m, l, o, p[g + 14], 23, -35309556);
        o = md5_hh(o, n, m, l, p[g + 1], 4, -1530992060);
        l = md5_hh(l, o, n, m, p[g + 4], 11, 1272893353);
        m = md5_hh(m, l, o, n, p[g + 7], 16, -155497632);
        n = md5_hh(n, m, l, o, p[g + 10], 23, -1094730640);
        o = md5_hh(o, n, m, l, p[g + 13], 4, 681279174);
        l = md5_hh(l, o, n, m, p[g + 0], 11, -358537222);
        m = md5_hh(m, l, o, n, p[g + 3], 16, -722881979);
        n = md5_hh(n, m, l, o, p[g + 6], 23, 76029189);
        o = md5_hh(o, n, m, l, p[g + 9], 4, -640364487);
        l = md5_hh(l, o, n, m, p[g + 12], 11, -421815835);
        m = md5_hh(m, l, o, n, p[g + 15], 16, 530742520);
        n = md5_hh(n, m, l, o, p[g + 2], 23, -995338651);
        o = md5_ii(o, n, m, l, p[g + 0], 6, -198630844);
        l = md5_ii(l, o, n, m, p[g + 7], 10, 11261161415);
        m = md5_ii(m, l, o, n, p[g + 14], 15, -1416354905);
        n = md5_ii(n, m, l, o, p[g + 5], 21, -57434055);
        o = md5_ii(o, n, m, l, p[g + 12], 6, 1700485571);
        l = md5_ii(l, o, n, m, p[g + 3], 10, -1894446606);
        m = md5_ii(m, l, o, n, p[g + 10], 15, -1051523);
        n = md5_ii(n, m, l, o, p[g + 1], 21, -2054922799);
        o = md5_ii(o, n, m, l, p[g + 8], 6, 1873313359);
        l = md5_ii(l, o, n, m, p[g + 15], 10, -30611744);
        m = md5_ii(m, l, o, n, p[g + 6], 15, -1560198380);
        n = md5_ii(n, m, l, o, p[g + 13], 21, 1309151649);
        o = md5_ii(o, n, m, l, p[g + 4], 6, -145523070);
        l = md5_ii(l, o, n, m, p[g + 11], 10, -1120210379);
        m = md5_ii(m, l, o, n, p[g + 2], 15, 718787259);
        n = md5_ii(n, m, l, o, p[g + 9], 21, -343485551);
        o = safe_add(o, j);
        n = safe_add(n, h);
        m = safe_add(m, f);
        l = safe_add(l, e)
    }
    return Array(o, n, m, l)
}

function md5_cmn(h, e, d, c, g, f) {
    return safe_add(bit_rol(safe_add(safe_add(e, h), safe_add(c, f)), g), d)
}

function md5_ff(g, f, k, j, e, i, h) {
    return md5_cmn((f & k) | ((~f) & j), g, f, e, i, h)
}

function md5_gg(g, f, k, j, e, i, h) {
    return md5_cmn((f & j) | (k & (~j)), g, f, e, i, h)
}

function md5_hh(g, f, k, j, e, i, h) {
    return md5_cmn(f ^ k ^ j, g, f, e, i, h)
}

function md5_ii(g, f, k, j, e, i, h) {
    return md5_cmn(k ^ (f | (~j)), g, f, e, i, h)
}

function core_hmac_md5(c, f) {
    var e = str2binl(c);
    if (e.length > 16) {
        e = core_md5(e, c.length * chrsz)
    }
    var a = Array(16),
        d = Array(16);
    for (var b = 0; b < 16; b++) {
        a[b] = e[b] ^ 909522486;
        d[b] = e[b] ^ 1549556828
    }
    var g = core_md5(a.concat(str2binl(f)), 512 + f.length * chrsz);
    return core_md5(d.concat(g), 512 + 128)
}

function safe_add(a, d) {
    var c = (a & 65535) + (d & 65535);
    var b = (a >> 16) + (d >> 16) + (c >> 16);
    return (b << 16) | (c & 65535)
}

function bit_rol(a, b) {
    return (a << b) | (a >>> (32 - b))
}

function str2binl(d) {
    var c = Array();
    var a = (1 << chrsz) - 1;
    for (var b = 0; b < d.length * chrsz; b += chrsz) {
        c[b >> 5] |= (d.charCodeAt(b / chrsz) & a) << (b % 32)
    }
    return c
}

function binl2str(c) {
    var d = "";
    var a = (1 << chrsz) - 1;
    for (var b = 0; b < c.length * 32; b += chrsz) {
        d += String.fromCharCode((c[b >> 5] >>> (b % 32)) & a)
    }
    return d
}

function binl2hex(c) {
    var b = hexcase ? "0123456789ABCDEF" : "0123456789abcdef";
    var d = "";
    for (var a = 0; a < c.length * 4; a++) {
        d += b.charAt((c[a >> 2] >> ((a % 4) * 8 + 4)) & 15) + b.charAt((c[a >> 2] >> ((a % 4) * 8)) & 15)
    }
    return d
}

function binl2b64(d) {
    var c = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
    var f = "";
    for (var b = 0; b < d.length * 4; b += 3) {
        var e = (((d[b >> 2] >> 8 * (b % 4)) & 255) << 16) | (((d[b + 1 >> 2] >> 8 * ((b + 1) % 4)) & 255) << 8) | ((d[
            b + 2 >> 2] >> 8 * ((b + 2) % 4)) & 255);
        for (var a = 0; a < 4; a++) {
            if (b * 8 + a * 6 > d.length * 32) {
                f += b64pad
            } else {
                f += c.charAt((e >> 6 * (3 - a)) & 63)
            }
        }
    }
    return f
};

对比一下本地和网页的运行结果，结果一致，说明抠出来的代码是正确的

代码整理

import requests
import time
import subprocess


yuanrenxue_headers = {
    "User-Agent": "yuanrenxue.project",
    "sessionid": "ihybaybtfp3htl48jf5mk4kx5bj7xbk3",
}


def get_m():
    ts_13 = int(time.time()) * 1000 + 100000000
    process = subprocess.run(f"node 第一关.js {str(ts_13)}", stdout=subprocess.PIPE)
    _result = process.stdout  # 获得标准输出
    m = _result.decode().strip()
    return m + "丨" + str(int(ts_13 / 1000))


def main():
    for page in range(1, 6):
        params = {"page": page, "m": get_m()}
        price_api = "http://match.yuanrenxue.com/api/match/1"
        resp = requests.get(price_api, params=params, headers=yuanrenxue_headers)
        # print(resp.request.url)
        print(resp.json()['data'])


if __name__ == "__main__":
    main()

注意是 “丨” ，不是 “|” （被这玩意卡了一下午 😇）

第一关.js 主要是在文件结尾补充了

// 接受时间戳参数
var arguments = process.argv.splice(2);
var ts_13 = arguments[0];
console.log(hex_md5(ts_13))

运行结果

猿人学 第一关：js 混淆 - 源码乱码

代码整理

猿人学第一关：js 混淆 - 源码乱码