C#实现百度网站收录和排名查询功能思路及实例

/// <summary>
        /// 服务提供者
        /// </summary>
        protected String SearchProvider { get; set; }

        /// <summary>
        /// 需要检索的网址
        /// </summary>
        protected String SiteUrl { get; set; }

        /// <summary>
        /// 搜索服务提供网址
        /// </summary>
        protected String BaseUrl { get; set; }

        /// <summary>
        /// 后页面网址
        /// </summary>
        /// <param name="site">需要查询的网址</param>
        /// <returns>拼接后的网址</returns>
        protected String GetDownUrl(string site)
        {
            return string.Format(BaseUrl, HttpUtility.UrlEncode(site));
        }

/// <summary>
        /// 查询在该搜索引擎中的收录次数
        /// </summary>
        /// <param name="siteurl">网站URL</param>
        public void SearchIncludeCount(string siteurl)
        {
            SiteUrl = siteurl;
            WebClient client = new WebClient();
            client.Encoding = Encoding.UTF8;
            client.DownloadStringCompleted += DownloadStringCompleted;
            client.DownloadStringAsync(new Uri(GetDownUrl(siteurl)));
        }

        /// <summary>
        /// 检索收录次数的具体实现
        /// 子类必须要实现该方法
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        protected abstract void DownloadStringCompleted(object sender, DownloadStringCompletedEventArgs e);

/// <summary>
        /// 当完成一个网站的收录查询后回调
        /// </summary>
        public Action<SiteIncludeCountResult> OnComplatedOneSite { get; set; }

/// <summary>
    /// 用于网站收录中委托的参数
    /// </summary>
    public class SiteIncludeCountResult
    {
        /// <summary>
        /// 收录次数
        /// </summary>
        public long IncludeCount { get; set; }

        /// <summary>
        /// 搜索引擎类型
        /// </summary>
        public String SearchType { get; set; }

        /// <summary>
        /// 网站URL
        /// </summary>
        public String SiteUrl { get; set; }
}

最后还有一个方法用于DownloadStringCompleted完成后回调OnComplatedOneSite委托：
        /// <summary>
        /// 完成处理后调用该方法将结果返回
        /// </summary>
        /// <param name="result">网址的收录数结果</param>
        protected void SetCompleted(SiteIncludeCountResult result)
        {
            if (OnComplatedOneSite != null)
                OnComplatedOneSite(result);
        }

百度为您找到相关结果约([\w,]+?)个

/// <summary>
    /// 百度网站收录次数查询
    /// </summary>
    public class BaiDuSiteIncludeCount : SiteIncludeCountBase
    {
        public BaiDuSiteIncludeCount()
        {
            BaseUrl = "http://www.baidu.com/s?wd=site%3A{0}";
            SearchProvider = "百度";
        }

        protected override void DownloadStringCompleted(object sender, DownloadStringCompletedEventArgs e)
        {
            var result = new SiteIncludeCountResult();
            result.SiteUrl = SiteUrl;
            result.SearchType = SearchProvider;
            result.IncludeCount = 0;
            Regex reg = new Regex(@"百度为您找到相关结果约([\w,]+?)个", RegexOptions.IgnoreCase | RegexOptions.Singleline);
            var matchs = reg.Matches(e.Result);
            if (matchs.Count > 0)
            {
                string count = matchs[0].Groups[1].Value.Replace(",", "");
                result.IncludeCount = long.Parse(count);
            }
            SetCompleted(result);
        }
}

/// <summary>
    /// 实现关键词查询必须继承该类
    /// </summary>
    public abstract class KeyWordsSeoBase
    {
        protected String BaseUrl { get; set; }

        protected String SearchProvider { get; set; }

        protected String GetDownUrl(string keyword, string site, long current)
        {
            return String.Format(BaseUrl, HttpUtility.UrlEncode(keyword), current);
        }

        protected void SetCompleted(KeyWordsSeoResult result)
        {
            if (OnComplatedOneKeyWord != null)
            {
                OnComplatedOneKeyWord(result);
            }
        }

        /// <summary>
        /// 完成一个关键词的查询后回调该委托
        /// </summary>
        public Action<KeyWordsSeoResult> OnComplatedOneKeyWord { get; set; }

        /// <summary>
        /// 查询指定关键词和网站在该搜索引擎中的排行
        /// 子类需要重写该方法
        /// </summary>
        /// <param name="keywords">关键词</param>
        /// <param name="site">网站URL</param>
        public abstract void SearchRanking(IEnumerable<string> keywords, string site,long count);
}

/// <summary>
    /// 用于关键词排行查询的委托参数
    /// </summary>
    public class KeyWordsSeoResult
    {
        /// <summary>
        /// 搜索引擎类型
        /// </summary>
        public String SearchType { get; set; }

        /// <summary>
        /// 关键词
        /// </summary>
        public String KeyWord { get; set; }

        /// <summary>
        /// 排行
        /// </summary>
        public long Ranking { get; set; }
    }

<span\s+class=""(?:g|c-showurl)"">([^/&]*)

protected KeyWordsSeoResult SearchFunc(string key, string siteurl, long total)
        {
            var result = new KeyWordsSeoResult();
            result.KeyWord = key;
            result.Ranking = total + 1;
            var reg = new Regex(@"<span\s+class=""(?:g|c-showurl)"">([^/&]*)", RegexOptions.IgnoreCase | RegexOptions.Singleline);
            var replace = new Regex("</?b>", RegexOptions.IgnoreCase | RegexOptions.Singleline);
            var client = new WebClient();
            long current = 0;
            long pos = 0;
            for (; ; )
            {
                String url = GetDownUrl(key, siteurl, current);
                String downstr = client.DownloadString(url);
                downstr = replace.Replace(downstr, "");
                var matchs = reg.Matches(downstr);
                foreach (Match match in matchs)
                {
                    pos++;
                    string suburl = match.Groups[1].Value;
                    try
                    {
                        if (suburl.ToLower() == siteurl.ToLower())
                        {
                            result.Ranking = pos;
                            return result;
                        }
                    }
                    catch
                    {
                        continue;
                    }
                }
                current += 10;
                if (current > total)
                {
                    current -= 10;
                    if (current >= total)
                    {
                        break;
                    }
                    current = total;
                }
            }
            return result;
        }

 public override void SearchRanking(IEnumerable<string> keywords, string site, long count)
        {
            new Thread(() =>
            {
                foreach (string key in keywords)
                {
                    KeyWordsSeoResult result = SearchFunc(key, site, count);
                    result.SearchType = SearchProvider;
                    SetCompleted(result);
                }
            }).Start();
        }
 

/// <summary>
    /// 查询网站的收录次数以及排行
    /// </summary>
    public class RankingAndIncludeSeo
    {
        /// <summary>
        /// 关键词列表
        /// </summary>
        public IList<KeyWordsSeoBase> KeyWordsSeoList { get; private set; }
       
        /// <summary>
        /// 收录次数列表
        /// </summary>
        public IList<SiteIncludeCountBase> SiteIncludeCountList { get; private set; }

        public RankingAndIncludeSeo()
        {
            KeyWordsSeoList = new List<KeyWordsSeoBase>();
            SiteIncludeCountList = new List<SiteIncludeCountBase>();
        }

        /// <summary>
        /// 当完成一个关键词的查询后回调该委托
        /// </summary>
        public Action<KeyWordsSeoResult> OnComplatedAnyKeyWordsSearch { get; set; }

        /// <summary>
        /// 当完成一个网站的收录次数查询后回调该委托
        /// </summary>
        public Action<SiteIncludeCountResult> OnComplatedAnySiteIncludeSearch { get; set; }

        /// <summary>
        /// 查询网址的排行
        /// </summary>
        /// <param name="keywords">关键词组</param>
        /// <param name="siteurl">查询的网址</param>
        /// <param name="count">最大限制排行数</param>
        public void SearchKeyWordsRanking(IEnumerable<string> keywords, string siteurl, long count = 100)
        {
            if (keywords == null)
                throw new ArgumentNullException("keywords", "必须存在关键词");
            if (siteurl == null)
                throw new ArgumentNullException("siteurl", "必须存在网站URL");
            foreach (KeyWordsSeoBase kwsb in KeyWordsSeoList)
            {
                kwsb.OnComplatedOneKeyWord = kwsb.OnComplatedOneKeyWord ?? OnComplatedAnyKeyWordsSearch;
                kwsb.SearchRanking(keywords, siteurl, count);
            }
        }

        /// <summary>
        /// 查询网址的收录次数
        /// </summary>
        /// <param name="siteurl">查询的网址</param>
        public void SearchSiteIncludeCount(string siteurl)
        {
            if (siteurl == null)
                throw new ArgumentNullException("siteurl", "必须指定网站");
            foreach (SiteIncludeCountBase sicb in SiteIncludeCountList)
            {
                sicb.OnComplatedOneSite = sicb.OnComplatedOneSite ?? OnComplatedAnySiteIncludeSearch;
                sicb.SearchIncludeCount(siteurl);
            }
        }
}