用微软控件执行网页中的脚本：msscript.ocx - 布颜书

用到的控件有： msscript.ocx ， msscrchs.dll 。引用其中的 msscript.ocx 在项目中即可；

示例代码如下：

代码

        #region 搜索歌曲的下载歌曲
        /// <summary>
        /// 从百度的试听页面获取歌曲的下载地址
        /// </summary>
        /// <param name="url"></param>
        /// <returns></returns>
        public static string DecodeMP3Url(string url)
        {
            string content = GetReponseText(url, Encoding.GetEncoding("gb2312"), string.Empty);

            Regex objRegex = new Regex("var sertim =(?<sertimvalue>[\\S ]*?), statim =(?<statimvalue>[\\S ]*?);",
                   RegexOptions.IgnoreCase | RegexOptions.Compiled);
            Match objMatch = objRegex.Match(content);

            if (objMatch.Success)
            {
                //获取变量定义
                string vardefine = objMatch.Value;

                objRegex = new Regex("function decode\\(url\\){(?<body>[\\S\\s]*?)return decurl;(?<g>[\\s]*?)}",
                   RegexOptions.IgnoreCase | RegexOptions.Compiled);
                objMatch = objRegex.Match(content);
                if (objMatch.Success)
                {
                    //获取加密的函数体
                    string functiondefine = objMatch.Value;

                    //获取加密后的下载地址列表
                    objRegex = new Regex("\"(?<t>[\\S]{4})://(?<g>[\\S]*?)\"", RegexOptions.IgnoreCase | RegexOptions.Compiled);
                    objMatch = objRegex.Match(content);
                    List<string> pathlist = new List<string>();
                    while (objMatch.Success)
                    {
                        string v = objMatch.Value.Substring(1, objMatch.Value.Length - 2);
                        if (!v.Contains("...")
                            && !pathlist.Contains(v)
                            && !v.StartsWith("http://", StringComparison.InvariantCultureIgnoreCase))
                        {
                            pathlist.Add(v);
                        }
                        objMatch = objMatch.NextMatch();
                    }

                    ScriptControl sc = new ScriptControl();
                    sc.Language = "JavaScript";
                    sc.AllowUI = false;
                    sc.AddCode(vardefine);
                    sc.AddCode(functiondefine);

                    foreach (string path in pathlist)
                    {
                        //解密
                        string realurl = sc.Eval(string.Format("decode('{0}')", path)).ToString();
                        if (CheckMusicUrl(realurl))
                        {
                            return realurl;
                        }
                    }
                }
            }

            return string.Empty;
        }
        static string GetReponseText(string sUrl, Encoding encode, string sAgent)
        {
            HttpWebRequest request = HttpWebRequest.Create(sUrl) as HttpWebRequest;
            //request.AllowAutoRedirect = false;
            Uri uri = new Uri(sUrl);
            request.Referer = "http://" + uri.Host;
            request.UserAgent = sAgent;

            string result;
            WebResponse response = request.GetResponse();
            using (Stream stream = response.GetResponseStream())
            {
                using (StreamReader reader = new StreamReader(stream, encode))
                {
                    result = reader.ReadToEnd();
                }
            }
            response.Close();
            return result;
        }

上面的代码是采集百度的MP3音乐，因为百度的音乐URL是用JS脚本构造出来的

所以我就直接用他的脚本，在C#后台进行URL构造，然后抓取地址内容。

posted on 2010-02-22 11:59 布颜书阅读(1783) 评论(0) 收藏举报

刷新页面返回顶部