lucene全文检索应用示例及代码简析

  Lucene是apache软件基金会 jakarta项目组的一个子项目,是一个开放源代码的全文检索引擎工具包及架构,提供了完整的查询引擎和索引引擎,实现了一些通用的分词算法,预留很多词法分析器接口。本文以 myrss.easyjf.com网站系统中使用Lucene实现全文检索的代码为例,简单演示Lucene在实际项目中的应用。
  使用Lucene实现全文检索,主要有下面三个步骤:
  1、建立索引库:根据网站新闻信息库中的已有的数据资料建立Lucene索引文件。
  2、通过索引库搜索:有了索引后,即可使用标准的词法分析器或直接的词法分析器实现进行全文检索。
  3、维护索引库:网站新闻信息库中的信息会不断的变动,包括新增、修改及删除等,这些信息的变动都需要进一步反映到Lucene索引文件中。
    下面是myrss.easyjf.com相关代码!
 
 一、索引管理(建立及维护)
  索引管理类MyRssIndexManage主要实现根据网站信息库中的数据建立索引,维护索引等。由于索引的过程需要消耗一定的时间,因此,索引管理类实现Runnable接口,使得我们可以在程序中开新线程来运行。
package com.easyjf.lucene;
import java.util.Date;
import java.util.List;
import org.apache.lucene.analysis.standard.StandardAnalyzer;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.Field;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.queryParser.MultiFieldQueryParser;
import org.apache.lucene.queryParser.QueryParser;
import org.apache.lucene.search.Hits;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.Searcher;
import com.easyjf.dbo.EasyJDB;
import com.easyjf.news.business.NewsDir;
import com.easyjf.news.business.NewsDoc;
import com.easyjf.news.business.NewsUtil;
import com.easyjf.web.tools.IPageList;
public class MyRssIndexManage implements Runnable {
 private String indexDir;
 private String indexType="add"; 
 public void run() {
  // TODO Auto-generated method stub
  if("add".equals(indexType))
   normalIndex();
  else if ("init".equals(indexType)) reIndexAll();
 }
 public void normalIndex()
 {
  try{
   Date start = new Date();
   int num=0;
   IndexWriter writer=new IndexWriter(indexDir,new StandardAnalyzer(),false);   
   //NewsDir dir=NewsDir.readBySn();   
   String scope="(needIndex<2) or(needIndex is null)";
   IPageList pList=NewsUtil.pageList(scope,1,50);
   for(int p=0;p<pList.getPages();p++)
   {
   pList=NewsUtil.pageList(scope,p,100);
   List list=pList.getResult();
   for(int i=0;i<list.size();i++)
   {
    NewsDoc doc=(NewsDoc)list.get(i);     
    writer.addDocument(newsdoc2lucenedoc(doc));   
    num++;
   }
   }
   writer.optimize();
   writer.close();
   EasyJDB.getInstance().execute("update NewsDoc set needIndex=2 where "+scope);
   Date end = new Date();
   System.out.print("新增索引"+num+"条信息,一共花:"+(end.getTime() - start.getTime())/60000+"分钟!");  
   }
   catch(Exception e)
   {
    e.printStackTrace();
   }
 }
 public void reIndexAll()
 {
  try{
   Date start = new Date();
   int num=0;
   IndexWriter writer=new IndexWriter(indexDir,new StandardAnalyzer(),true);   
   NewsDir dir=NewsDir.readBySn("easyjf");   
   IPageList pList=NewsUtil.pageList(dir,1,50);
   for(int p=0;p<pList.getPages();p++)
   {
   pList=NewsUtil.pageList(dir,p,100);
   List list=pList.getResult();
   for(int i=0;i<list.size();i++)
   {    
    NewsDoc doc=(NewsDoc)list.get(i);     
    writer.addDocument(newsdoc2lucenedoc(doc));
    num++;
   }
   }
   writer.optimize();
   writer.close();
   EasyJDB.getInstance().execute("update NewsDoc set needIndex=2 where dirPath like 'easyjf%'");
   Date end = new Date();
   System.out.print("全部重新做了一次索引,一共处理了"+num+"条信息,花:"+(end.getTime() - start.getTime())/60000+"分钟!");  
   }
   catch(Exception e)
   {
    e.printStackTrace();
   }
 }
 private Document newsdoc2lucenedoc(NewsDoc doc)
 {
  Document lDoc=new Document();     
  lDoc.add(new Field("title",doc.getTitle(),Field.Store.YES,Field.Index.TOKENIZED));
  lDoc.add(new Field("content",doc.getContent(),Field.Store.YES,Field.Index.TOKENIZED));
  lDoc.add(new Field("url",doc.getRemark(),Field.Store.YES,Field.Index.NO));    
  lDoc.add(new Field("cid",doc.getCid(),Field.Store.YES,Field.Index.NO));
  lDoc.add(new Field("source",doc.getSource(),Field.Store.YES,Field.Index.NO));
  lDoc.add(new Field("inputTime",doc.getInputTime().toString(),Field.Store.YES,Field.Index.NO));
  return lDoc;
 }
 public String getIndexDir() {
  return indexDir;
 }
 public void setIndexDir(String indexDir) {
  this.indexDir = indexDir;
 }
 
 public String getIndexType() {
  return indexType;
 }
 public void setIndexType(String indexType) {
  this.indexType = indexType;
 }
}
 
二、使用Lucene实现全文搜索
   下面是MyRssSearch类的源码,该类主要实现使用Lucene中Searcher及QueryParser实现从索引库中搜索关键词。
package com.easyjf.lucene;
import java.util.List;
import org.apache.lucene.analysis.standard.StandardAnalyzer;
import org.apache.lucene.document.Document;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.queryParser.MultiFieldQueryParser;
import org.apache.lucene.queryParser.QueryParser;
import org.apache.lucene.search.Hits;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.Searcher;
import com.easyjf.search.MyRssUtil;
import com.easyjf.search.SearchContent;
import com.easyjf.web.tools.IPageList;
import com.easyjf.web.tools.PageList;
public class MyRssSearch {
 private String indexDir; 
 IndexReader ir;
 Searcher search;
 public IPageList search(String key,int pageSize,int currentPage)
 {
  IPageList pList=new PageList(new HitsQuery(doSearch(key)));
  pList.doList(pageSize,currentPage,"","",null);
  if(pList!=null)
  {  
   List list=pList.getResult(); 
   if(list!=null){
   for(int i=0;i<list.size();i++)
   {
    list.set(i,lucene2searchObj((Document)list.get(i),key));
   }
   }
  }
  try{
  if(search!=null)search.close();
  if(ir!=null)ir.close();
  }
  catch(Exception e)
  {
   e.printStackTrace();
  }
  return pList;
 }
 private SearchContent lucene2searchObj(Document doc,String key)
 {
  SearchContent searchObj=new SearchContent();
  String title=doc.getField("title").stringValue();
  searchObj.setTitle(title.replaceAll(key,"<font color=red>"+key+"</font>"));  
  searchObj.setTvalue(doc.getField("cid").stringValue());  
  searchObj.setUrl(doc.getField("url").stringValue());
  searchObj.setSource(doc.getField("source").stringValue());  
  searchObj.setLastUpdated(doc.getField("inputTime").stringValue()); 
  searchObj.setIntro(MyRssUtil.content2intro(doc.getField("content").stringValue(),key));  
  return searchObj;
 }
 public Hits doSearch(String key)
 {
  Hits hits=null;
  try{
  ir=IndexReader.open(indexDir);
  search=new IndexSearcher(ir);  
  String fields[]={"title","content"};
  QueryParser parser=new MultiFieldQueryParser(fields,new StandardAnalyzer());
  Query query=parser.parse(key);
  hits=search.search(query);   
  }
  catch(Exception e)
  {
   e.printStackTrace();
  }
  //System.out.println("搜索结果:"+hits.length());
  return hits;
 }
 
 public String getIndexDir() {
  return indexDir;
 }
 public void setIndexDir(String indexDir) {
  this.indexDir = indexDir;
 }
}

 
  在上面的代码中,search方法返回一个封装了分页查询结果的IPageList,IPageList是EasyJWeb Tools业务引擎中的分页引擎,对于IPageList的使用,请看本人写的这篇文章《 EasyJWeb Tools中业务引擎分页的设计实现》:

  我们针对Lucene的的查询结果Hits结构,写了一个查询器HitsQuery。代码如下所示:
package com.easyjf.lucene;
import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
import org.apache.lucene.search.Hits;
import com.easyjf.web.tools.IQuery;
public class HitsQuery implements IQuery {
 private int begin=0;
 private int max=0;
 private Hits hits;
 public HitsQuery()
 {
  
 }
 public HitsQuery(Hits hits)
 {
  if(hits!=null)
  {
   this.hits=hits;   
   this.max=hits.length();
  }
 }
 public int getRows(String arg0) {
  // TODO Auto-generated method stub
  return (hits==null?0:hits.length());
 }
 public List getResult(String arg0) {
  // TODO Auto-generated method stub
  List list=new ArrayList();  
  for(int i=begin;i<(begin+max)&&(i<hits.length());i++)
  {
   try{
   list.add(hits.doc(i));
   }
   catch(Exception e)
   {
    e.printStackTrace();
   }
  }
  return list;
 }
 public void setFirstResult(int begin) {
  // TODO Auto-generated method stub
  this.begin=begin;
 }
 public void setMaxResults(int max) {
  // TODO Auto-generated method stub
  this.max=max;
 }
 public void setParaValues(Collection arg0) {
  // TODO Auto-generated method stub
  
 }
 public List getResult(String condition, int begin, int max) {
  // TODO Auto-generated method stub
  if((begin>=0)&&(begin<max))this.begin=begin;
  if(!(max>hits.length()))this.max=max;
  return getResult(condition);
 }
}
  三、Web调用
  下面我们来看看在Web中如果调用商业逻辑层的全文检索功能。下面是处理用户请请的Action中关于搜索部分的源码:
package com.easyjf.news.action;
public class SearchAction implements IWebAction { 
public Page doSearch(WebForm form,Module module)throws Exception
{
 String key=CommUtil.null2String(form.get("v")); 
 key=URLDecoder.decode(URLEncoder.encode(key,"ISO8859_1"),"utf-8");
 form.set("v",key);
 form.addResult("v2",URLEncoder.encode(key,"utf-8"));
 if(key.getBytes().length>2){
 String orderBy=CommUtil.null2String(form.get("order")); 
 int currentPage=CommUtil.null2Int(form.get("page"));
 int pageSize=CommUtil.null2Int(form.get("pageSize"));  
 if(currentPage<1)currentPage=1;
 if(pageSize<1)pageSize=15;
 SearchEngine search=new SearchEngine(key,orderBy,pageSize,currentPage);
 search.getLuceneSearch().setIndexDir(Globals.APP_BASE_DIR+"/WEB-INF/index");
 search.doSearchByLucene();
 IPageList pList=search.getResult();
 if(pList!=null && pList.getRowCount()>0){
  form.addResult("list",pList.getResult());
  form.addResult("pages",new Integer(pList.getPages()));
  form.addResult("rows",new Integer(pList.getRowCount()));
  form.addResult("page",new Integer(pList.getCurrentPage()));
  form.addResult("gotoPageHTML",CommUtil.showPageHtml(pList.getCurrentPage(),pList.getPages()));
  }
 else 
 {
  form.addResult("notFound","true");//找不到数据
 } 
 }
 else
  form.addResult("errMsg","您输入的关键字太短!");
 form.addResult("hotSearch",SearchEngine.getHotSearch(20));
 return null;
}
}
 
其中调用的SearchEngine类中有关Lucene部分的源码:
public class SearchEngine {
private MyRssSearch luceneSearch=new MyRssSearch();
public void doSearchByLucene()
{ 
 SearchKey keyObj=readCache(); 
 if(keyObj!=null){
  result=luceneSearch.search(key,pageSize,currentPage);  
  if(updateStatus){
  keyObj.setReadTimes(new Integer(keyObj.getReadTimes().intValue()+1));
  keyObj.update();
  }  
 }
 else//缓存中没有该关键字信息,生成关键字搜索结果
 { 
  keyObj=new SearchKey();
  keyObj.setTitle(key);
  keyObj.setLastUpdated(new Date());
  keyObj.setReadTimes(new Integer(1));
  keyObj.setStatus(new Integer(0));
  keyObj.setSequence(new Integer(1));
  keyObj.setVdate(new Date());
  keyObj.save(); 
  result=luceneSearch.search(key,pageSize,currentPage);; 
  
 } 
}
}
 
四、程序演示效果
  这是EasyJF团队官方网站上提供java信息搜索的 myrss.easyjf.com的运行效果。
     lucene全文检索应用示例及代码简析_第1张图片

你可能感兴趣的:(apache,搜索引擎,Lucene,webform,全文检索)