lucene.net 搜索及分页

lucene.net 搜索及分页2008-10-21 23:20整理下lucene.net 搜索及分页代码(可扩展为可缓存，有快照的小型搜索引擎) 已测

using System;
using System.Collections;
using System.ComponentModel;
using System.Data;
using System.Data.SqlClient;
using System.Drawing;
using System.Web;
using System.Web.SessionState;
using System.Web.UI;
using System.Web.UI.WebControls;
using System.Web.UI.HtmlControls;
using Lucene.Net;
using Lucene.Net.Index;
using Lucene.Net.Documents;


using Lucene.Net.QueryParsers;
using Lucene.Net.Search;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.Analysis.Cn;
using Lucene.Net.Highlight;
using System.IO;
using Lucene.Net.Analysis;
using System.Configuration;
using System.Text.RegularExpressions;


namespace HTWJWeb_System.HTWJSearch
{
    public partial class HTWJSearch : BaseSearch
    {


        #region 处理搜索并将信息转换为可显示结果数据源
        /**/
        /// <summary>
        /// Does the search and stores the information about the results.
        /// </summary>

        public void search()
        {


            // 索引目录
            //string indexDirectory = Server.MapPath(ConfigurationSettings.AppSettings["EnableCache"] );
            //创建一个Searcher用于搜索
            //记录查询开始的时间

            DateTime start = DateTime.Now;
            this.GetIndexDir("HTWJ", IndexDiectory);
            IndexSearcher searcher = new IndexSearcher(dir);

            //从"FullText"字段搜索
            //Console.WriteLine(this.Query);

            ChineseAnalyzer OneAnalyzer = new ChineseAnalyzer();
            BooleanClause.Occur[] flags = new BooleanClause.Occur[] {
                   BooleanClause.Occur.SHOULD, BooleanClause.Occur.SHOULD };

           // Query query = QueryParser.Parse(this.Query, "FullText", new StandardAnalyzer());

            //多字段检索
            Query query = MultiFieldQueryParser.Parse(this.Query, new String[] { "Title", "FullText" }, flags, OneAnalyzer);


            //Query query = parser.Parse(this.Query);

            //创建结果记录集
            //定义字段
            this.Results.Columns.Add("File_Id", typeof(string));
            this.Results.Columns.Add("Title", typeof(string));
            this.Results.Columns.Add("FullText", typeof(string));

            Sort sort = new Sort(new SortField("Title", SortField.DOC, true));

         
            //Hits是搜索结果记录集,不过是Lucene自己的格式,需要格式化成标准输出
            Hits hits = searcher.Search(query, sort);
            //结果个数
            this.total = hits.Length();
            /**/
            ////创建高亮显示
            //Highlighter highlighter = new Highlighter(new SimpleHTMLFormatter("<font color="#C60A00">", "</font>"), new QueryScorer(query));
            //highlighter.TextFragmenter = new SimpleFragmenter(160);
            //highlighter.MaxDocBytesToAnalyze = 256;
            // initialize startAt

            ////创建高亮显示2
            SimpleHTMLFormatter simpleHTMLFormatter = new SimpleHTMLFormatter("<B><font color=\"red\">", "</font></B>");
            Highlighter highlighter = new Highlighter(simpleHTMLFormatter, new QueryScorer(query));

            this.startAt = initStartAt();
            // how many items we should show - less than defined at the end of the results
            int resultsCount = smallerOf(total, this.maxResults + this.startAt);
            for (int i = startAt; i < resultsCount; i++)
            {
                Document doc = hits.Doc(i);
                DataRow row = this.Results.NewRow();
                row["File_Id"] = doc.Get("File_Id");

                //后期处理---链接
                row["Title"] = "<a href=\"http://www.zzu.edu.cn\" target=\"_blank\"> " + doc.Get("Title") + "</a>";
                
                //ChineseAnalyzer analyzer1 = new ChineseAnalyzer();
                TokenStream tokenStream = OneAnalyzer.TokenStream("FullText", new StringReader(doc.Get("FullText")));
                row["FullText"] = highlighter.GetBestFragment(tokenStream, doc.Get("FullText"));

                //用这种方法可直接创建高亮 可扩展为高亮链接
                //string fullText = doc.Get("FullText");
                //row["FullText"] = this.SimpleHighLighter(fullText, this.Query, "<B><font color=\"red\">", "</font></B>", 126);

                this.Results.Rows.Add(row);
            }


            searcher.Close();
            // result information


            this.fromItem = startAt + 1;
            this.toItem = smallerOf(startAt + maxResults, total);
            //记录查询使用的时间
            this.duration = DateTime.Now - start;
        }
        #endregion


        #region 页面上面信息
        /**/
        /// <summary>
        /// 页面上面信息
        /// </summary>

        public DataTable pageinfo
        {
            get 
            {
                DataTable da = new DataTable();
                da.Columns.Add("info", typeof(string));
                DataRow bar = da.NewRow();
                bar["info"] = "总共有" + this.total + "项, 用时" + this.duration + " 当前从" + this.fromItem + " 到" + this.toItem + " ";
                da.Rows.Add(bar);
                return da;
            }
        }
        #endregion


        #region 页面底航连接
        /**/
        /// <summary>
        /// 页面底航连接
        /// </summary>
        public DataTable Paging
        {
            get
           {
                int pageNumber = (startAt + maxResults - 1) / maxResults;

                DataTable dt = new DataTable();
                dt.Columns.Add("html", typeof(string));

                //增加第一页链接
                DataRow tar = dt.NewRow();
                if (startAt >= maxResults)
                    tar["html"] = "<EM><a href=\"/HTWJSearch.aspx?q=" + Server.UrlEncode(this.Query) + "&start=" + (startAt - maxResults) + "\"><IMG src=\"images/b_pre.gif\"></a></EM>";
                else
                    tar["html"] = "<EM><IMG src=\"images/b_pre.gif\"></EM>";
                dt.Rows.Add(tar);


                int previousPagesCount = 7;

                DataRow ar = dt.NewRow();
                ar["html"] = pagingItemHtml(startAt, pageNumber + 1, false);
                dt.Rows.Add(ar);

                for (int i = pageNumber - 1; i >= 0 && i >= pageNumber - previousPagesCount; i--)
                {
                    int step = i - pageNumber;
                    DataRow r = dt.NewRow();
                    r["html"] = pagingItemHtml(startAt + (maxResults * step), i + 1, true);

                    dt.Rows.InsertAt(r, 1);
                }

                int nextPagesCount = 8;
                for (int i = pageNumber + 1; i <= pageCount && i <= pageNumber + nextPagesCount; i++)
                {
                    int step = i - pageNumber;
                    DataRow r = dt.NewRow();
                    r["html"] = pagingItemHtml(startAt + (maxResults * step), i + 1, true);

                    dt.Rows.Add(r);
                }

                //增加第一页链接
                DataRow far = dt.NewRow();
                if (pageNumber < pageCount)
                    far["html"] = "<EM><a class=\"blue1\" href=\"/HTWJSearch.aspx?q=" + Server.UrlEncode(this.Query) + "&start=" + (startAt + maxResults) + "\"><IMG src=\"images/b_nextpage.gif\"></a></EM>";
                else
                    far["html"] = "<EM><IMG src=\"images/b_nextpage.gif\"></EM>";
                dt.Rows.Add(far);
                return dt;
            }
        }


        #region 页面连接列表

        /**//// <summary>
        /// 页面连接列表
        /// </summary>
        /// <param name="start">开始</param>
        /// <param name="number">显示数量</param>
        /// <param name="active">活动</param>
        /// <returns></returns>
        public string pagingItemHtml(int start, int number, bool active)
        {

            if (active)
                return "<VAR><a href=\"/HTWJSearch.aspx?q=" + Server.UrlEncode(this.Query) + "&start=" + start + "\">" + number + "</a></VAR>";
            else
                return "<VAR class=on>" + number + "</VAR>";
        }
        #endregion


        #endregion


        #region This should be replaced with a direct client-side get
        /// <summary>
        /// This should be replaced with a direct client-side get
        /// </summary>
        /// <param name="sender"></param>
        /// <param name="e"></param>
        protected void ButtonSearch_Click(object sender, System.EventArgs e)
        {
            this.Response.Redirect("HTWJSearch.aspx?q=" + this.TextBoxQuery.Text.Trim());
        }

        #endregion




            #region Web Form Designer generated code
        override protected void OnInit(EventArgs e)
        {
            //
            // CODEGEN: This call is required by the ASP.NET Web Form Designer.
            //
            InitializeComponent();
            base.OnInit(e);
        }

        /// <summary>
        /// Required method for Designer support - do not modify
        /// the contents of this method with the code editor.
        /// </summary>
        private void InitializeComponent()
        {

        }
        #endregion


    }

}
 

lucene.net 搜索及分页22008-10-21 23:21//父类

using System;
using System.Collections;
using System.ComponentModel;
using System.Data;
using System.Data.SqlClient;
using System.Drawing;
using System.Web;
using System.Web.SessionState;
using System.Web.UI;
using System.Web.UI.WebControls;
using System.Web.UI.HtmlControls;
using Lucene.Net;
using Lucene.Net.Index;
using Lucene.Net.Documents;


using Lucene.Net.QueryParsers;
using Lucene.Net.Search;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.Analysis.Cn;
using Lucene.Net.Highlight;
using System.IO;
using Lucene.Net.Analysis;


namespace HTWJWeb_System
{
    public partial class BaseSearch : System.Web.UI.Page
    {

        #region 变量声明
        /**/
        /// <summary>
        /// 搜索结果数据
        /// </summary>
        public DataTable Results = new DataTable();
        /**/
        /// <summary>
        /// 开始索引
        /// </summary>
        public int startAt;
        /**/
        /// <summary>
        /// First item on page (user format).
        /// </summary>
        public int fromItem;
        /**/
        /// <summary>
        /// Last item on page (user format).
        /// </summary>
        public int toItem;
        /**/
        /// <summary>
        /// 搜索的结果总数
        /// </summary>
        public int total;
        /**/
        /// <summary>
        /// 搜索所用时间
        /// </summary>
        public TimeSpan duration;
        /**/
        /// <summary>
        /// 每页显示结果项条数
        /// </summary>
        public int maxResults = 3;
        /**/
        /// <summary>
        /// 是否启用网页缓存功能
        /// </summary>
        public bool EnableCache;
        /**/
        /// <summary>
        /// 缓存URL
        /// </summary>
        public string CacheURL;
        /**/
        /// <summary>
        /// 索引文件存放的路径
        /// </summary>
        public string IndexDiectory;
        private string m_Query;
        public Lucene.Net.Store.Directory dir;
        #endregion 变量声明



        #region 取得查询目标索引的缓存

        public void GetIndexDir(string IndexKey, string IndexDiectory)
        {
            object obj = Cache[IndexKey];
            if (obj != null)
            {
             dir = (Lucene.Net.Store.Directory)obj;
          }
           else
            {
              dir = new Lucene.Net.Store.RAMDirectory(IndexDiectory);
               //Cache.Insert(IndexKey, dir, null, DateTime.Now.AddMinutes(2), TimeSpan.Zero);
            }
            dir = new Lucene.Net.Store.RAMDirectory(IndexDiectory);
        }
        #endregion


        #region 得到定长的字符串
        /**/
        /// <summary>
        /// 得到定长的字符串
        /// </summary>
        /// <param name="p_Text">原字符串</param>
        /// <param name="p_Length">长度</param>
        /// <param name="p_ExtraText">多余部分显示字符</param>
        /// <returns></returns>
        public string GetLengthText(string p_Text, int p_Length, string p_ExtraText)
        {
            return (p_Text.Length > p_Length) ? (p_Text.Substring(0, 45) + p_ExtraText) : p_Text;
        }
        #endregion

        #region 取得两个参数中的最小值
        /**/
        /// <summary>
        /// 取得两个参数中的最小值
        /// </summary>
        /// <param name="first">参数一</param>
        /// <param name="second">参数二</param>
        /// <returns>最小值</returns>
        public int smallerOf(int first, int second)
        {
            return first < second ? first : second;
        }
        #endregion


        #region 检测开始位置
        /**/
        /// <summary>
        /// Initializes startAt value. Checks for bad values.
        /// </summary>
        /// <returns></returns>
        public int initStartAt()
        {
            try
            {
                int sa = Convert.ToInt32(this.Request.Params["start"]);
                // too small starting item, return first page
                if (sa < 0)
                    return 0;
                // too big starting item, return last page
                if (sa >= total - 1)
                {
                    return lastPageStartsAt;
                }
                return sa;
            }
            catch
            {
                return 0;
            }
        }
        #endregion

        #region 最后一页的第一项
        /**/
        /// <summary>
        /// First item of the last page
        /// </summary>
        public int lastPageStartsAt
        {
            get
            {
                return pageCount * maxResults;
            }
        }


        public int pageCount
        {
            get
            {
                return (total - 1) / maxResults; // floor
            }
        }
        #endregion

        #region 取得高亮显示
        /**/
        /// <summary>
        /// 取得高亮连接
        /// </summary>
        /// <param name="p_Body">处理内容</param>
        /// <param name="p_KeyWords">关键词</param>
        /// <returns></returns>
        public string SimpleHighLighter(string p_Body, string p_KeyWords, string p_Before, string p_After, int p_MaxLength)
        {
            string[] KeyWords = p_KeyWords.Trim().Split(' ');
            if (p_Body.Length > p_MaxLength)
           {
            if (p_Body.IndexOf(KeyWords[0]) > 10)
               {
                    try
                   {
                        if ((p_Body.Length - 10) > p_MaxLength)
                            p_Body = p_Body.Substring(p_Body.IndexOf(KeyWords[0]) - 10, p_MaxLength) + "";
                        else
                            p_Body = p_Body.Substring(p_Body.IndexOf(KeyWords[0]) - 10) + "";
                    }
                    catch
                   { }
              }
               else
                    p_Body = p_Body.Substring(0, p_MaxLength) + "";
           }
            
            for (int i = 0; i < KeyWords.Length; i++)
            {
                p_Body = p_Body.Replace(KeyWords[i], p_Before + KeyWords[i] + p_After);
            }
            return p_Body;
        }
        #endregion


        #region
        private string SimpleHighLightHelper(string p_Body, string p_KeyWords)
        {
            return "";

        }
        #endregion

        #region 属性
        /**/
        /// <summary>
        /// 查询关键词
        /// </summary>
        public string Query
        {
            get
            {
                return m_Query;
            }
            set
            {
                m_Query = value;
            }
        }
        #endregion


     
    }
}