搜索引擎

开发平台：
C#

ClassCaiYang.cs：源码内容
							using System;
using System.IO;
using System.Collections.Generic;
using System.Text;
using System.Collections;
/*
      '       迅龙中文分类搜索引擎  v0.6
      '
      '        LGPL  许可发行
      '
      '       宁夏大学  张冬 康彩  zd4004@163.com
      ' 
      '        官网 http://blog.163.com/zd4004/
 */
namespace XunLong.binAutoModelTeam
{
    class ClassCaiYang
    {
        /// <summary>
        /// 目录列表队列
        /// </summary>
        ArrayList LIST = new ArrayList();
        /// <summary>
        /// 文件系统对象
        /// </summary>
        NetHashTableAPI.ClassNHT db = new NetHashTableAPI.ClassNHT();
        
        /// <summary>
        ///  文件保存路径
        /// </summary>
       public  string savePath;
        /// <summary>
        /// 读取各个模版  记录和存储 第一个采样的url 和 模板名称(由原始模板数据MD5生成)
        /// </summary>
        public void Init(string filePath, string savePathX, string k_c_path)
        {
            db.SetClassNHT(filePath, 3145727, k_c_path);
            LIST = db.SearchOneList("http:");
            Console.WriteLine(" 共有 ==>> " + LIST.Count.ToString()+"  条数据");
            savePath = savePathX;
        }
        public void Run()
        {
            int BN = 0;
            if (System.IO.File.Exists(savePath) == true)
            {
                System.IO.File.Delete(savePath);
            }
            while (LIST.Count > 0)
            {
                //取出第一个
                string a_one = LIST[0].ToString();
                LIST.Remove(a_one);
                // key = url  val  =  相似度  >12
                Hashtable TMP_H = new Hashtable();
               
                //建模的等级 从高到低               
                    for (int i = 0; i < LIST.Count; i++)
                    {
                        int n1 = XunLong.UrlStringLib.ClassUrlString.Url2Url(a_one, LIST[i].ToString());
                        if (n1 >= 12) 
                        {
                            TMP_H.Add(LIST[i],n1);
                        }
                    }
                    for (int u = 50; u >= 12; u--)
                    {
                        ArrayList TMP = new ArrayList();
                        TMP.Clear();
                        foreach (System.Collections.DictionaryEntry de in TMP_H)
                        {
                            if ((int)de.Value >= u)
                            {
                                TMP.Add(de.Key);
                            }
                        }
                        if (TMP.Count >=64)  //个数大于３６　　等级最高的
                        {
                            foreach (string b_a in TMP)
                            {
                                LIST.Remove(b_a);
                                TMP_H.Remove(b_a);
                            }
                            //写入一条数据
                            putmOLDMODELSOURCEFileData(savePath, a_one);
                            BN = BN + 1;
                            Console.WriteLine("-建模等级->" + u.ToString());
                            Console.WriteLine("-共识别出可模版化数据-> " + BN.ToString() + " 条");
                            goto x_NEXT;       //只取符合条件　等级最高的
                        }
                    }
                    if (LIST.Count % 10 == 0)
                    {
                        Console.Write(" " + LIST.Count.ToString());
                    }
                x_NEXT: ;
            }
            Console.WriteLine("识别完毕");
            Console.WriteLine("数据保存在: ");
            Console.WriteLine(savePath);
        }
        /// <summary>
        /// 写入已经使用过的一条数据　　　xxx-->> 相对于设置列表 olgurl
        /// </summary>
        /// <param name="filename">文件名</param>
        /// <param name="data">数据</param>
        /// <param name="isApp">是否追加模式</param>
        public void putmOLDMODELSOURCEFileData(string okPath, string data)
        {
            StreamWriter writer = null;
            try
            {
                writer = new StreamWriter(okPath, true, System.Text.Encoding.GetEncoding("gb2312"));
                //  writer.Write(data);
                writer.WriteLine(data);
                writer.Close();
            }
            catch (IOException e)
            {
                Console.WriteLine(e.Message);
            }
            finally
            {
                if (writer != null)
                    writer.Close();
            }
        }
    }
}

						