初试Lucene.net搜索及高亮分页-CSDN博客

Code
using System;
using System.Data;
using System.IO;
using System.Text.RegularExpressions;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.Documents;
using Lucene.Net.QueryParsers;
using Lucene.Net.Search;
using Lucene.Net.Search.Highlight;
using Lucene.Net.Analysis;
using Lucene.Net.Index;
using System.Configuration;

namespace So

{
public class BaseSearch : System.Web.UI.Page

{
 变量声明#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 = 10;

 /**//// <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

 取得查询目标索引的缓存#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

 得到定长的字符串#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

 取得两个参数中的最小值#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

 检测开始位置#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

 最后一页的第一项#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

 取得高亮连接#region 取得高亮连接

 /**//// <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

 属性#region 属性
 /**//// <summary>
 /// 查询关键词
 /// </summary>
 public string Query

{
get

{
                return m_Query;
            }
            set

{
                m_Query = value;
            }
        }
        #endregion
    }
}

using System;
using System.Data;
using System.IO;
using System.Text.RegularExpressions;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.Documents;
using Lucene.Net.QueryParsers;
using Lucene.Net.Search;
using Lucene.Net.Search.Highlight;
using Lucene.Net.Analysis;
using Lucene.Net.Index;
using System.Configuration;

namespace So.News

{
public class NewsSearch : BaseSearch

{
public NewsSearch()

{
 this.IndexDiectory = ConfigurationManager.AppSettings["NewsIndexPath"];
 }

 处理搜索并将信息转换为可显示结果数据源#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("HDC.News", IndexDiectory);
 IndexSearcher searcher = new IndexSearcher(dir);
 //从"body"字段搜索
 //Console.WriteLine(this.Query);

 Lucene.Net.Analysis.Analyzer OneAnalyzer = new StandardAnalyzer();
 QueryParser parser = new QueryParser("newsContent", OneAnalyzer);
 Query query = parser.Parse(this.Query);

 //创建结果记录集
 //定义字段
 this.Results.Columns.Add("ArticleID", typeof(int));
 this.Results.Columns.Add("ArticleClassID", typeof(int));
 this.Results.Columns.Add("className", typeof(string));
 this.Results.Columns.Add("titleImg", typeof(string));
 this.Results.Columns.Add("updateTime", typeof(DateTime));
 this.Results.Columns.Add("source", typeof(string));
 this.Results.Columns.Add("title", typeof(string));
 this.Results.Columns.Add("summary", typeof(string));

 Sort sort = new Sort(new SortField("ArticleID", 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("", ""), new QueryScorer(query));
 //highlighter.TextFragmenter = new SimpleFragmenter(160);
 //highlighter.MaxDocBytesToAnalyze = 256;

 // initialize startAt
 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["ArticleID"] = Convert.ToInt32(doc.Get("ArticleID"));
 row["ArticleClassID"] = Convert.ToInt32(doc.Get("ArticleClassID"));
 string summary = doc.Get("summary");

 row["summary"] = this.SimpleHighLighter(summary, this.Query,"", "",226);
 row["className"] = doc.Get("className");
 row["titleImg"] = doc.Get("titleImg");
 row["updateTime"] = Convert.ToDateTime(doc.Get("updateTime"));
 row["source"] = doc.Get("source");
 row["title"] = doc.Get("title");
 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

 页面底航连接#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"] = "<a href="/News/?q=" + Server.UrlEncode(this.Query) + "&start=" + (startAt - maxResults) + ""><IMG src="images/b_pre.gif"></a>";
 else
 tar["html"] = "<IMG src="images/b_pre.gif">";
 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"] = "<a class="blue1" href="/News/?q=" + Server.UrlEncode(this.Query) + "&start=" + (startAt + maxResults) + ""><IMG src="images/b_nextpage.gif"></a>";
 else
 far["html"] = "<IMG src="images/b_nextpage.gif">";
 dt.Rows.Add(far);
 return dt;
 }
 }

 页面连接列表#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="/News/?q=" + Server.UrlEncode(this.Query) + "&start=" + start + "">" + number + "</a></VAR>";
 else
 return "<VAR class=on>" + number + "</VAR>";
 }
 #endregion

 #endregion

 }
}

分类: .NET

标签: Lucene.net, 搜索

本文转自孤独侠客博客园博客，原文链接：http://www.cnblogs.com/lonely7345/archive/2008/06/27/1231354.html，如需转载请自行联系原作者