数据采集基础案例-JS分析（一）

public void BuildMap(String[] oriWords, String[] targetWords)
{
     var currentWord = String.Empty; ;
    for (Int32 i = 0; i < oriWords.Length; ++i)
    {
        currentWord = oriWords[i];
        var num = Int32.Parse(currentWord);
        num =num%2>0?num+1:num-1;
        targetWords[i] = Char.ConvertFromUtf32(num);
    }
}

public interface ISymbolTable
{
    Boolean IsReady { get; }
    Task<Boolean> ReadyAsync(CollectContextInfo context);
    Task<String> GetSymbolAsync(String key, CollectContextInfo context);
}

public class CollectContextInfo
{
    public Uri Uri { get; set; }
    public String Original { get; set; }
    public ISymbolTable SymbolTable { get; set; }
    /// <summary>
    /// 表示当前原始内容解析的定位点
    /// </summary>
    public Int32 CurrentPosition { get; set; }
    public CollectResult Result { get; set; }

    public String[] KeywordsTable { get; set; }
}

public class CollectResult
{
    public String Title { get; set; }
    public String Content { get; set; }
    public String PreUri { get; set; }
    public String NextUri { get; set; }
}

  public class WebBroswerSymbolTable : ISymbolTable
  {
    //...
  }

    HtmlElement ele = this._wbBroswer.Document.CreateElement("script");
    ele.SetAttribute("type", "text/javascript");
    ele.SetAttribute("text", @"
    function getCssText() 
    { 
        var rulesObj=document.styleSheets[0].rules;
        var result=''; 
        for(var i in rulesObj)
        {    
          if(rulesObj[i].style)
          {
             result+=('|'+rulesObj[i].selectorText+'='+rulesObj[i].style.cssText);
           } 
         } 
         return result;
     }");
     this._wbBroswer.Document.Body.AppendChild(ele);

  public class DefaultSymbolTable : ISymbolTable
  {
  ///...
  }

            //获取用于分析的 JS 代码
            Regex jsRegex = new Regex(@"(?<=u.AES=p._createHelper\(d\)}\)\(\);)[\s\S]+(?=</script>)");
            var jsMatch = jsRegex.Match(mapDataStr);
            if (!jsMatch.Success)
            {
                return isReady;
            }
            var jsStr = jsMatch.Value;

            //获取 strData 信息
            Regex strDataRegex = new Regex(@"(?<=var[\s\S]+\[)[\s\S]+?(?=];)");
            var strDataMatch = strDataRegex.Match(jsStr);
            if (!strDataMatch.Success)
            {
                return isReady;
            }
            Int32 startIndex = strDataMatch.Index + strDataMatch.Length;

            var strDataArray = strDataMatch.Value.Split(',');
            var strDataList = new List<String>();
            for (Int32 i = 0; i < strDataArray.Length; ++i)
            {
                strDataArray[i] = strDataArray[i].Trim('\'');
                strDataList.Add(strDataArray[i]);
            }

            //循环将第一个元素移动至末尾
            //获取循环次数
            Regex loopCountRegex = new Regex(@"(?<=push[\s\S]+,)[\s\S]+?(?=\)\))");
            var loopCountMatch = loopCountRegex.Match(jsStr, startIndex);
            if (!loopCountMatch.Success)
            {
                return isReady;
            }
            startIndex = loopCountMatch.Index + loopCountMatch.Value.Length;
            var loopCount = Convert.ToInt32(loopCountMatch.Value, 16);
            while ((--loopCount) >= 0)
            {
                var fisrt = strDataList.First();
                strDataList.RemoveAt(0);
                strDataList.Add(fisrt);
            }
            strDataArray = strDataList.ToArray();

            //获取 data 在字符数组中的索引     var data = _0xea12('0xc');
            Regex dataIndexRegex = new Regex(@"(?<=var data=.*\')[\s\S]+?(?=\'.*;)");
           var dataIndexMatch= dataIndexRegex.Match(jsStr, startIndex);
            if (!dataIndexMatch.Success)
            {
                return isReady;
            }
            startIndex = dataIndexMatch.Index + dataIndexMatch.Value.Length;

            var dataStr = dataIndexMatch.Value;
            if (dataStr.StartsWith("0x"))
            {
                var dataIndex = Convert.ToInt32(dataStr, 16);
                dataStr = strDataArray[dataIndex];
            }
           


            //获取 Key 字符串  var keywords = CryptoJS[_0xea12('0xd')][_0xea12('0xe')][_0xea12('0xf')]('6B0600CA9BCE5B24');
            Regex keyIndexRegex = new Regex(@"(?<=var keywords=CryptoJS.+]\(.*\')[\s\S]+?(?=\'.*\);)");
            var keyIndexMatch = keyIndexRegex.Match(jsStr, startIndex);
            if (!keyIndexMatch.Success)
            {
                return isReady;
            }
            startIndex = keyIndexMatch.Index + keyIndexMatch.Value.Length;
            var keyIndexStr = keyIndexMatch.Value;
            var keyStr = keyIndexStr;
            if (keyIndexStr.StartsWith("0x"))
            {
                var keyIndex = Convert.ToInt32(keyIndexStr, 16);
                keyStr = strDataArray[keyIndex];
            }

            //获取 IV 字符串
            Regex ivIndexRegex = new Regex(@"(?<=iv=CryptoJS.+]\(.*\')[\s\S]+?(?=\'.*\);)");
            var ivIndexMatch = ivIndexRegex.Match(jsStr, startIndex);
            if (!ivIndexMatch.Success)
            {
                return isReady;
            }
            startIndex = ivIndexMatch.Index + ivIndexMatch.Value.Length;
            var ivIndexStr = ivIndexMatch.Value;
            var ivStr = ivIndexStr;
            if (ivIndexStr.StartsWith("0x"))
            {
                var ivIndex = Convert.ToInt32(ivIndexStr, 16);
                ivStr = strDataArray[ivIndex];
            }

            //解密数据
            Encoding iso = Encoding.GetEncoding("ISO-8859-1");
            Byte[] key = iso.GetBytes(keyStr);
            Byte[] iv = iso.GetBytes(ivStr);

            var decryMapDataStr = String.Empty;
            using (var aesDecry = new AesManaged())
            {
                aesDecry.Padding = PaddingMode.Zeros;
                aesDecry.Mode = CipherMode.CBC;
                aesDecry.Key = key;
                aesDecry.IV = iv;
                ICryptoTransform decryptor = aesDecry.CreateDecryptor(aesDecry.Key, aesDecry.IV);
                var oriDataByte = Convert.FromBase64String(dataStr);

                using (MemoryStream msDecrypt = new MemoryStream(oriDataByte))
                {
                    using (CryptoStream csDecrypt = new CryptoStream(msDecrypt, decryptor, CryptoStreamMode.Read))
                    {
                        using (StreamReader srDecrypt = new StreamReader(csDecrypt))
                        {
                            decryMapDataStr = srDecrypt.ReadToEnd();
                        }
                    }
                }

                decryMapDataStr = decryMapDataStr.TrimEnd('\0');
                var oriWords = decryMapDataStr.Split(',');
                var targetWords = new String[oriWords.Length];
                BuildMap(oriWords, targetWords);
                //....
            }

//....