会飞的鱼

FS - RM
谁能反抗生命随波逐流的离合
首页 » 综合整理 » 某站视频真实地址解析过程

某站视频真实地址解析过程


前不久看到一篇关于BibiBili视频地址解析的源码,正好有兴趣,就自己也研究了研究,记录一下。

1.以http://www.bilibili.com/video/av12535537/此视频为例用HttpWatch抓包,很容易就能看出视频的地址是从https://interface.bilibili.com/playurl?quality=4&player=1&cid=20627240&ts=1504780011&sign=f37df88434579a4b009a2bee708a2a71&qn=112里获取的
按照常例分析下参数:
cid很容易猜测出是视频的ID号,在网页源码里面也能找到

QQ图片20170916125103.jpg

Quality和qn是视频的清晰度
ts则是时间戳,这个也不用多说
关键就是这个sign,在HttpWatch中是完全搜不到任何相关的信息的,但网页访问了http://static.hdslb.com/play.swf这个文件,那么sign算法很可能就在这个播放器里。

2.下载这个SWF文件后,载入AS3 Sorcerer,代码一大片不是很好找


fsgsgsd.jpg



那就保存到文本里,在文本中查找”视频地址”这个关键字。


pic_proxy_.jpg

很快就能定位到关键的函数LoadPreview和LoadCidVideo那么接着搜索LoadPreview和LoadCidVideo定位到函数部分,可以看到代码如下

pic_pr.jpg

pic.jpg

那么关键算法就是在getSign和getSign_v2里面了


public function getSign(_arg_1:String):String
    {
        var _local_2:int;
        var _local_4 = null;
        var _local_7:int;
        var _local_8:int;
        var _local_6:int;
        var _local_5:int;
        var _local_3:int = ESP;
        _local_2 = _local_3;
        _local_3 = (_local_3 - 48);
        _local_5 = 16;
        _local_6 = (_local_2 - 37);
        ESP = (_local_3 & -16);
        _local_7 = CModule.mallocString(_arg_1);
        _local_8 = _arg_1.length;
        do
        {
            var _local_9:int = (L__2E_str2 - _local_5);
            _local_9 = li8((_local_9 + 16)) /*FlasCC (Alchemy)*/ ;
            _local_3 = (_local_3 - 16);
            si32(_local_9, (_local_3 + 4)); //FlasCC (Alchemy)
            si32(_local_6, _local_3); //FlasCC (Alchemy)
            ESP = _local_3;
            F_sprintf();
            _local_3 = (_local_3 + 16);
            _local_5 = (_local_5 + -1);
            _local_6 = (_local_6 + 2);
        } while (_local_5 != 0);
        _local_3 = (_local_3 - 16);
        _local_9 = (_local_2 - 4);
        si32(_local_9, (_local_3 + 12)); //FlasCC (Alchemy)
        _local_9 = (_local_2 - 37);
        si32(_local_9, (_local_3 + 8)); //FlasCC (Alchemy)
        si32(_local_8, (_local_3 + 4)); //FlasCC (Alchemy)
        si32(_local_7, _local_3); //FlasCC (Alchemy)
        ESP = _local_3;
        F_get_sign();
        _local_3 = (_local_3 + 16);
        _local_5 = eax;
        _local_9 = li32((_local_2 - 4)) /*FlasCC (Alchemy)*/ ;
        _local_4 = CModule.readString(_local_5, _local_9);
        if (_local_7 != 0)
        {
            _local_3 = (_local_3 - 16);
            si32(_local_7, _local_3); //FlasCC (Alchemy)
            ESP = _local_3;
            F_idalloc();
            _local_3 = (_local_3 + 16);
        };
        if (_local_5 != 0)
        {
            _local_3 = (_local_3 - 16);
            si32(_local_5, _local_3); //FlasCC (Alchemy)
            ESP = _local_3;
            F_idalloc();
            _local_3 = (_local_3 + 16);
        };
        var _local_10 = _local_4;
        _local_3 = _local_2;
        ESP = _local_3;
        return (_local_10);
    }
}//package com.bilibili.interfaces
华丽的分割线-------



public function getSign_v2(_arg_1:String, _arg_2:int):String
    {
        var _local_12:*;
        var _local_3:int;
        var _local_5 = null;
        var _local_6:int;
        var _local_11:int;
        var _local_8:int;
        var _local_10:int;
        var _local_9:int;
        var _local_4:int = ESP;
        _local_3 = _local_4;
        _local_4 = (_local_4 - 48);
        _local_6 = _arg_2;
        if (_local_6 >= 5)
        {
            _local_12 = _arg_1;
        }
        else
        {
            _local_8 = (L__2E_str2 + (_local_6 << 4));
            _local_9 = 16;
            _local_10 = (_local_3 - 37);
            ESP = (_local_4 & -16);
            _local_6 = CModule.mallocString(_arg_1);
            _local_11 = _arg_1.length;
            do
            {
                var _local_7:int = (_local_8 - _local_9);
                _local_7 = li8((_local_7 + 16)) /*FlasCC (Alchemy)*/ ;
                _local_4 = (_local_4 - 16);
                si32(_local_7, (_local_4 + 4)); //FlasCC (Alchemy)
                si32(_local_10, _local_4); //FlasCC (Alchemy)
                ESP = _local_4;
                F_sprintf();
                _local_4 = (_local_4 + 16);
                _local_9 = (_local_9 + -1);
                _local_10 = (_local_10 + 2);
            } while (_local_9 != 0);
            _local_4 = (_local_4 - 16);
            _local_7 = (_local_3 - 4);
            si32(_local_7, (_local_4 + 12)); //FlasCC (Alchemy)
            _local_7 = (_local_3 - 37);
            si32(_local_7, (_local_4 + 8)); //FlasCC (Alchemy)
            si32(_local_11, (_local_4 + 4)); //FlasCC (Alchemy)
            si32(_local_6, _local_4); //FlasCC (Alchemy)
            ESP = _local_4;
            F_get_sign();
            _local_4 = (_local_4 + 16);
            _local_10 = eax;
            _local_7 = li32((_local_3 - 4)) /*FlasCC (Alchemy)*/ ;
            _local_5 = CModule.readString(_local_10, _local_7);
            if (_local_6 != 0)
            {
                _local_4 = (_local_4 - 16);
                si32(_local_6, _local_4); //FlasCC (Alchemy)
                ESP = _local_4;
                F_idalloc();
                _local_4 = (_local_4 + 16);
            };
            if (_local_10 != 0)
            {
                _local_4 = (_local_4 - 16);
                si32(_local_10, _local_4); //FlasCC (Alchemy)
                ESP = _local_4;
                F_idalloc();
                _local_4 = (_local_4 + 16);
            };
            _local_12 = _local_5;
        };
        _local_4 = _local_3;
        ESP = _local_4;
        return (_local_12);
    }
}//package com.bilibili.interfaces


可以看到,二者代码非常相似,而且光看代码无法找到有用的信息,那么就得想办法去调试了

3.目前我只知道的办法是:使用JPEXS反编译软件来修改SWF代码(插Log),然后通过Fiddler劫持替换SWF,通过显示Log信息来进行调试

要显示log信息就得安装debug版本的flash

http://www.adobe.com/support/flashplayer/debug_downloads.html

pic_proxy_meitu.jpg

为方便阅读日志信息可以安装Cygwin

pic_proxy_meitu_567678.jpg

具体步骤我也是参考了http://blog.csdn.net/hot_vc/article/details/50600717这篇文章

将SWF载入JPEXS

pic_proxy_meitu_6.jpg

要想修改SWF的源码就得去修改PCODE,第一次接触这种也不是很懂,觉得有点类似于C#的IL代码
通过AS3代码和PCODE代码的一一进行对比,还是能找到一些规律的
为了查看代码中变量值,用到以下代码

    findpropstrict Qname(PackageNamespace(""),"trace")
    getlocal 5
    callpropvoid Qname(PackageNamespace(""),"trace") 1

这个代码就相当于trace(_loc5_);
修改保存后利用Fiddler来劫持替换我们的SWF

pic_proxy_meitu_7.jpg

这样Cygwin就能输出我们想要看到的结果了。
其中有些变量的值是地址,而不是数据,为了读取地址中的数据,利用到了这句代码CModule.readString(地址,读取长度);
按照原PCODE的格式,写出插LOG时的PCODE

pushint 60
setlocal 8
getlex Qname(PackageNamespace("com.bilibili.interfaces"),"CModule")
getlocal 7
getlocal 8
callproperty Qname(PackageNamespace(""),"readString") 2
coerce_s
setlocal 5
findpropstrict Qname(PackageNamespace(""),"trace")
getlocal 5
callpropvoid Qname(PackageNamespace(""),"trace") 1

此代码相当于
_loc8_:int = 60;
_loc5_:* = CModule.readString(_loc7_,_loc8_);
trace(_loc5_);
pic_proxy_meitu_8.jpg


4.其实关键算法在F_get_sign()中,因为从F_get_sign()函数中出来后,eax中就已经存放带着sign的视频URL的地址了。


F_get_sign()函数过于复杂,总之通过不断地插这两种Log查看变量信息,最终找到了算法

pic_proxy_meitu_9.jpg

由这个地方可以看出是MD5,还有一个位置能够准确地查看到进行MD5处理的数据(已经忘了.....)
算法如下:

五种清晰度
qn=16,quality=1 
qn=32,quality=5
qn=48,quality=2
qn=64,quality=2
qn=80,quality=3
qn=112,quality=4
 
三种类型的视频bili2,bangumi,movie
 
bili2:
time = F_clock_gettime();
 
sign = MD5(“cid=” + cid + “&player=1&qn=112&quality=4&ts=” + time + “1c15888dc316e05a15fdd0a02ed6584f”)
 
url  =  “https://interface.bilibili.com/playurl?player=1&qn=112&cid=” + cid + “&quality=4&ts=” + time + “&sign=” + sign
 
bangumi:
time = F_clock_gettime();
 
sign = MD5(“cid=” + cid + “&module=bangumi” + “&player=1” + “&qn=112” + “&quality=4” + “&ts=” + time + “9b288147e5474dd2aa67085f716c560d”)
 
url  =  “https://bangumi.bilibili.com/player/web_api/playurl?cid=” + cid + “&player=1” + “&module=bangumi” + “&qn=112” + “&quality=4&ts=” + time + “&sign=” + sign

顺便说下,有些tx.acgvideo.com开头的视频无法下载是因为服务器检查了Refer,只要在协议头里面加入Refer:https://static.hdslb.com/play.swf,就能下载了




文章如无特别注明均为原创! 作者: 浮生若梦, 转载或复制请以 超链接形式 并注明出处 浮生若梦's Blog
原文地址《 某站视频真实地址解析过程》发布于2017-9-16

分享到:
0 打赏

评论

切换注册

登录

您也可以使用第三方帐号快捷登录

Q Q 登 录
微 博 登 录
切换登录

注册