猿人学Web端爬虫攻防大赛第十题 与某数ast+jsdom方法
(编辑:jimmy 日期: 2024/11/16 浏览:3 次 )
@[TOC](猿人学Web端爬虫攻防大赛第十题 与某数ast+jsdom方法)
参考文章:
1.猿人学爬虫攻防赛 第10题
2.某数和某5秒-反混淆动态注入调试的一种方案
样品网址:https://match.yuanrenxue.com/match/10
打开首页后直接就有5处混淆,第一处是某数套 jsjiami.com.v6,然后后面4处为obfuscator
首先还原这一段jsjiamiv6的混淆,还原后的结果如下
let yuanrenxue_36 = '';var yuanrenxue_59 = 968;for (let yuanrenxue_229 = 0; yuanrenxue_229 < $_ts["dfe1683"]["length"]; yuanrenxue_229++) {yuanrenxue_36 += String["fromCharCode"]($_ts["dfe1683"][yuanrenxue_229]["charCodeAt"]() - yuanrenxue_229 % yuanrenxue_59 - 50);}yuanrenxue_18 = atob(yuanrenxue_36);yuanrenxue_31(78, yuanrenxue_18);
可以大概看出其中的逻辑主要是对$_ts["dfe1683"]进行解密,接着还原rs的混淆
*******省略很多代码******* case 29: yuanrenxue_48.yuanrenxue_110 = new Date().getTime(); var yuanrenxue_35 = yuanrenxue_48["dfe1675"]; yuanrenxue_48["dfe1675"] = yuanrenxue_40; var yuanrenxue_91 = yuanrenxue_48.yuanrenxue_yuanrenxue_126; var yuanrenxue_13 = yuanrenxue_31(8); var yuanrenxue_21 = yuanrenxue_31(8); var yuanrenxue_15 = yuanrenxue_48.aebi = []; var yuanrenxue_88 = yuanrenxue_31(71); var yuanrenxue_65 = yuanrenxue_35.length; var yuanrenxue_1 = 0; var yuanrenxue_89 = yuanrenxue_12(); var yuanrenxue_54 = yuanrenxue_12(); var yuanrenxue_67 = yuanrenxue_12(); var yuanrenxue_72 = yuanrenxue_12(); var yuanrenxue_105 = yuanrenxue_12(); yuanrenxue_14 = yuanrenxue_12(); var yuanrenxue_28 = yuanrenxue_12(); if (yuanrenxue_14 > 0) { yuanrenxue_96 = yuanrenxue_35.substr(yuanrenxue_1, yuanrenxue_28).split(String.fromCharCode(255)); } yuanrenxue_1 += yuanrenxue_28; var yuanrenxue_56 = []; var yuanrenxue_14 = yuanrenxue_12(); for (yuanrenxue_24 = 0; yuanrenxue_24 < yuanrenxue_14; yuanrenxue_24++) { yuanrenxue_23(16, yuanrenxue_24, yuanrenxue_56); } for (yuanrenxue_24 = 0; yuanrenxue_24 < yuanrenxue_14; yuanrenxue_24++) { yuanrenxue_56.push("}"); } yuanrenxue_56.push(")();"); var yuanrenxue_18 = yuanrenxue_56.join(''); yuanrenxue_48.yuanrenxue_110 -= yuanrenxue_31(8); yuanrenxue_21 = yuanrenxue_31(8); let yuanrenxue_36 = ''; var yuanrenxue_59 = 479; for (let yuanrenxue_229 = 0x0; yuanrenxue_229 < $_ts["dfe1683"]["length"]; yuanrenxue_229++) { yuanrenxue_36 += String["fromCharCode"]($_ts["dfe1683"][yuanrenxue_229]["charCodeAt"]() - yuanrenxue_229 % yuanrenxue_59 - 0x32); } yuanrenxue_18 = atob(yuanrenxue_36); yuanrenxue_31(0x4e, yuanrenxue_18); var yuanrenxue_32 = yuanrenxue_31(8); if (yuanrenxue_32 - yuanrenxue_13 > 12000) { yuanrenxue_48.yuanrenxue_74 = 1; } return;*******省略很多代码*******
整个代码比较多,只贴上部分代码,可以看到["dfe1675"]解密后的代码赋值给了yuanrenxue_18 ,接着["dfe1683"]解密后的代码也是直接赋值给了yuanrenxue_18 ,相当于覆盖了,所以前面的一部分代码就不需要看了,根本没有执行。
接着后面还有4个ob混淆,但是里面的代码并不重要,只是反调试有点烦人,那么接着看yuanrenxue_18的代码,连续拿到两次yuanrenxue_18的代码,查看变化,发现有三处
这里都与后面的一个接口 https://match.yuanrenxue.com/api/offset 的返回值息息相关,说明这个值也是非常重要的。到这里还是没有找到加密参数生成的入口,但是请求发出的事情,参数就自动生成了,那么很有可能xhr下的open,send等方法被改写了
这就可以知道,参数是通过_yrxyA$函数生成的,使用ast在这个函数的结尾动态注入一个【console.log(_yrx2LR);】
这样当我们调用open方法的时候,就可以获取包含加密参数的最终链接,最后使用jsdom把代码跑起来,需要先安装依赖库
npm install jsdom
import requests_htmlimport jsonimport reimport base64import subprocessimport osimport logginglogging.disable(logging.INFO)def main(): requests = requests_html.HTMLSession() headers = { 'User-Agent': 'yuanrenxue.project', 'Referer': 'https://match.yuanrenxue.com/match/10', 'X-Requested-With': 'XMLHttpRequest' } base_url = 'https://match.yuanrenxue.com' top_url = base_url + '/match/10' response = requests.get(top_url, headers=headers) script = response.html.find('script') ts_1_url = base_url + script[0].attrs['src'] ts_2_url = base_url + script[1].attrs['src'] ts_3_url = base_url + '/api/offset' with open('config.json', 'w', encoding='utf-8') as f: f.write(json.dumps({ 'html': response.content.decode(), 'url': 'https://match.yuanrenxue.com/match/10' })) with open('html.html', 'wb') as f: f.write(response.content) yuanrenxue_59 = int(re.findall('(?<=var yuanrenxue_59=)\d+', response.content.decode())[0]) response = requests.get(ts_2_url, headers=headers) dfe1683 = response.content.decode()[17:-1] yuanrenxue_36 = '' for yuanrenxue_229 in range(len(dfe1683)): yuanrenxue_36 += chr(ord(dfe1683[yuanrenxue_229]) - yuanrenxue_229 % yuanrenxue_59 - 50) with open('eval.js', 'wb') as f: f.write(base64.b64decode(yuanrenxue_36.encode())) os.system('node ast_eval') os.system('node ast_html') with open('ts.js', 'wb') as f: response = requests.get(ts_1_url, headers=headers) f.write(response.content) with open('offset.js', 'wb') as f: response = requests.get(ts_3_url, headers=headers) f.write(response.content) with open('rs_decrypt.js', 'r', encoding='utf-8') as f: js = f.read() print('开始获取链接') for page in range(1, 6): with open('get_url.js', 'w', encoding='utf-8') as f: f.write("const{JSDOM}=require('jsdom');const fs = require('fs');const dom=new JSDOM(fs.readFileSync('html.html').toString('utf-8'),{url:'https://match.yuanrenxue.com/match/10'});window=dom.window;document=window.document;\n") f.write('if(typeof __dirname!="undefined"){__dirname=undefined}if(typeof __filename!="undefined"){__filename=undefined}if(typeof require!="undefined"){require=undefined}if(typeof exports!="undefined"){exports=undefined}if(typeof module!="undefined"){module=undefined}\n') f.write("window.eval.call(window, fs.readFileSync('ts.js').toString('utf-8') + fs.readFileSync('offset.js').toString('utf-8'));\n") f.write("window.$_ts = $_ts;\n") f.write(js) f.write('\n\nlet xml = new window.XMLHttpRequest();\n') f.write("""xml.open('GET', "/api/match/10?page=""" + str(page) + """\", false);\n""") f.write("process.exit(0);") command = ['node', 'get_url.js'] nodejs = subprocess.Popen(command, stderr=subprocess.PIPE, stdout=subprocess.PIPE) url = nodejs.stdout.read().decode().replace('\n', '') print(url) response = requests.get(url, headers=headers).json() print(response) with open('offset.js', 'rb') as f: offset = f.read() with open('offset.js', 'wb') as f: f.write(offset[:14] + response['k']['k'].split('|')[1].encode())if __name__ == '__main__': main()
这里因为每次请求接口都会返回一个新的offset,所以这里每次请求后都需要重新加载
非常好,可以成功请求5页的数据了。当然,这个方法运用在真实某数下也是可以的,例子如下
RS4代混淆
*******省略很多代码******* case 78: if (_$JX === undefined || _$JX === "") { return; } if (_$8k.execScript) { ret = _$8k.execScript(_$JX); return ret; } _$YA = _$8k.eval; ret = _$YA.call(_$8k, _$JX); return ret;*******省略很多代码*******
RS5代混淆
*******省略很多代码******* case 73: if (_$r1 === undefined || _$r1 === "") { return; } var _$_P; if (_$Z8["execScript"]) { _$_P = _$Z8["execScript"](_$r1); return _$_P; } _$RD = _$Z8["eval"]; _$_P = _$RD["call"](_$Z8, _$r1); return _$_P;*******省略很多代码*******
RS6代混淆
*******省略很多代码******* case 101: _$_r = undefined; if (_$hi === _$_r || _$hi === "") { return; } _$_r = _$fx; if (_$_r.execScript) { _$_n = _$_r.execScript(_$hi); return _$_n; } _$_d = _$_r.eval; _$_n = _$_d.call(_$_r, _$hi); return _$_n; case 41: _$_n = _$$X(38); _$_d = _$_z; _$cq = _$_d.nsd; _$$7 = _$by; _$_d.lcd = _$$7; _$_d.nsd = _$$7; _$cy = _$$X(114, _$cq); _$bb = _$$X(0, 711, _$$X(114, _$cq)); _$$d = []; _$_d.cp = _$$d; _$$d[1] = _$bb; _$aJ = _$_d.aebi = []; _$_d.scj = [{ "0": 0, "1": 1, "2": 2, "3": 3 }, { "0": 0, "1": 0, "2": 0, "3": 0, "4": 0 }, { "0": 0, "1": -545, "2": -538, "3": 174 }, { "0": 0, "1": 3, "2": 3, "3": 0, "4": 0, "5": -241, "6": -241 }]; _$d8 = "省略代码"; _$g$ = _$d8.length; _$_t = 0; _$$d[0] = "省略代码"; _$$d[2] = "省略代码"; _$$d[6] = ""; _$$r = _$_R(); _$$n = _$_R(); _$$b = _$_R(); _$er = _$_R(); _$dp = _$_R(); _$a5 = _$_R(); _$iH = _$_R() * 55295 + _$_R(); _$_d = _$bF; if (_$a5 > 0) { _$gj = _$d8.substr(_$_t, _$iH).split(_$_d.fromCharCode(257)); } _$_t += _$iH; _$a5 = _$_R(); _$aN = 0; for (; _$aN < _$a5;) { _$gj.push(_$$1(51, _$_R() * 55295 + _$_R())); _$aN++; } _$eP = []; _$_X = '\n\n\n\n\n'; _$a5 = _$_R(); _$aN = 0; for (; _$aN < _$a5;) { _$eP.push(_$_X.substr(0, _$cy() % 5)); _$$1(69, _$aN, _$eP); _$aN++; } _$eP.push('}}}}}}}}}}'.substr(_$a5 - 1)); _$$1(54, _$eP); _$eP.push("})($_ts.scj,$_ts.aebi);"); _$d5 = _$eP.join(''); _$$d[4] = _$$X(38) - _$_n; _$_f = 0; for (_$aN = 0; _$aN < _$d5.length; _$aN += 100) { _$_f += _$d5.charCodeAt(_$aN); } _$$d[3] = _$_f; _$_L = _$$X(38); _$$X(101, _$d5); _$$d[5] = _$$X(38) - _$_n; return; }*******省略很多代码*******
上一篇:中星睿典满分补丁制作过程
下一篇:记一次实习僧雪碧图字体加密反爬
几个月来,英特尔、微软、AMD和其它厂商都在共同推动“AI PC”的想法,朝着更多的AI功能迈进。在近日,英特尔在台北举行的开发者活动中,也宣布了关于AI PC加速计划、新的PC开发者计划和独立硬件供应商计划。
在此次发布会上,英特尔还发布了全新的全新的酷睿Ultra Meteor Lake NUC开发套件,以及联合微软等合作伙伴联合定义“AI PC”的定义标准。