searchengine.java

java 写的一个新闻发布系统

//
//                                   ____.
//                       __/\ ______|    |__/\.     _______
//            __   .____|    |       \   |    +----+       \
//    _______|  /--|    |    |    -   \  _    |    :    -   \_________
//   \\______: :---|    :    :           |    :    |         \________>
//           |__\---\_____________:______:    :____|____:_____\
//                                      /_____|
//
//                 . . . i n   j a h i a   w e   t r u s t . . .
//
//
//
//

package org.jahia.services.search;

import java.io.*;                       	// File access methods
import java.util.*;                     	// Vector

import org.apache.regexp.*;             	// Regular Expression

import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.index.IndexReader;
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 org.jahia.gui.*;             		// GuiBean
import org.jahia.utils.*;           		// JahiaConsole
import org.jahia.params.*;          		// ParamBean
import org.jahia.exceptions.*;
import org.jahia.data.search.*;     		// JahiaSearchResult
import org.jahia.registries.*;     	 		// JahiaRegistries
import org.jahia.data.fields.*;     		// JahiaFields
import org.jahia.services.pages.JahiaPage;
import org.jahia.services.usermanager.*; 	// JahiaUser


/**
 * Search engine using Lucene.
 *
 * @author Khue Nguyen <a href="mailto:[e-mail]">[e-mail]</a>
 */

class SearchEngine {

    private static final String CLASS_NAME = SearchEngine.class.getName();

	private Analyzer analyzer;
	private IndexReader reader;


	/**
	 * Constructor
	 *
	 * @param Analyzer, the analyzer.
	 * @param IndexReader reader, the index reader to use.
	 */
	public SearchEngine ( Analyzer analyzer, IndexReader reader )
	{
		this.analyzer = analyzer;
		this.reader = reader;
	}

	//--------------------------------------------------------------------------
	/**
     * Return a vector of matching pages.
     * Perform a search for a given query ( valid lucene query ).
     * The search is limited to the site returned by the param bean.
	 *
	 * @param JahiaSearcher jSearcher, the jahia searcher.
	 * @param ParamBean jParams, the param bean.
	 * @return JahiaSearchResult result, the search result as an JahiaSearchResult object.
	 */
	public JahiaSearchResult doPageSearch(	 JahiaSearcher jSearcher,
							   	 ParamBean jParams )
	throws JahiaException
	{
		JahiaConsole.println( CLASS_NAME+".doPageSearch", "Started" );

		JahiaSearchResult result = new JahiaSearchResult();

		if ( reader == null || jSearcher== null || jSearcher.getQuery()==null )
			return result;

		try {
			// instanciate the searcher with the site index.
			Searcher searcher = new IndexSearcher(reader);

			// instanciate the query
			Query query = QueryParser.parse(jSearcher.getQuery().toLowerCase(),
											JahiaSearchConstant.FIELD_VALUE,
											analyzer);
			// get the search result.
			Hits hits = searcher.search(query);
			result = buildJahiaSearchResult(hits,jParams);

		} catch ( Throwable t ) {
			JahiaConsole.println( CLASS_NAME+".doPageSearch",
								  "Exception : " + t.getMessage() );
		}
		return result;
	}

	//--------------------------------------------------------------------------
    /**
     * Return a vector of matching containers for a given container list.
     * Perform a search for a given query ( valid lucene query ).
     * The search is limited to the site returned by the param bean.
     *
	 * @param JahiaSearcher jSearcher, the jahia searcher.
	 * @param ParamBean jParams, the param bean.
	 * @return JahiaSearchResult result, the search result as an JahiaSearchResult object.
	 */
	public JahiaSearchResult doContainerSearch( JahiaSearcher jSearcher,
							   	 	 			ParamBean jParams )
    throws JahiaException
    {
		JahiaConsole.println( CLASS_NAME+".doContainerSearch", "Started" );

		JahiaSearchResult result = new JahiaSearchResult();

		if ( reader == null || jSearcher== null || jSearcher.getQuery()==null )
			return result;

		StringBuffer buff = new StringBuffer(1024);
		try {

			ContainerSearcher cSearcher = (ContainerSearcher)jSearcher;

			// instanciate the searcher with the site index.
			Searcher searcher = new IndexSearcher(reader);

			ServicesRegistry sReg = ServicesRegistry.getInstance();

			// restric result to containers of the ctnlist.
	        // get container list containers.
	        Vector ctnids = sReg.getJahiaContainersService().getctnidsInList( cSearcher.getCtnListID() );
			if ( ctnids.size() == 0 )
				return result;

			int nbCtn = ctnids.size();
			buff.append("(");
			buff.append(cSearcher.getQuery().toLowerCase());
			buff.append(") AND ( ");
			for ( int i=0 ; i<nbCtn ; i++ )
			{
				buff.append(JahiaSearchConstant.FIELD_CTNID);
				buff.append(":");
				buff.append(((Integer)ctnids.get(i)).toString());
				if ( i<(nbCtn-1) )
				{
					buff.append(" OR ");
				} else {
					buff.append(" ) ");
				}
			}

			JahiaConsole.println( CLASS_NAME+".doContainerSearch", "Query is : " + buff.toString() );

			// instanciate the query
			Query query = QueryParser.parse(buff.toString(),
											JahiaSearchConstant.FIELD_VALUE,
											this.analyzer);
			// get the search result.
			Hits hits = searcher.search(query);
			if ( hits == null || hits.length()==0 )
				return result;

			JahiaConsole.println( CLASS_NAME+".doContainerSearch", "lucene result : " + hits.length() );

			result = buildContainerSearchResult(hits,jParams);

			JahiaConsole.println( CLASS_NAME+".doContainerSearch", "jahia result : " + result.getHitCount() );

		} catch ( Throwable t ) {
			JahiaConsole.println( CLASS_NAME+".doContainerSearch",
								  "Exception : " + t.getMessage() );
		} finally {
			buff = null;
		}
        return result;
	}


	//--------------------------------------------------------------------------
	/**
	 * Build jahia search result ( matching container ids ) from hits returned by lucene.
	 * The vector results of the returned JahiaSearchResult is a vector of ids, not of containers !
	 * No right check on container, only on field.
	 *
	 * TODO : have a common method for building JahiaSearchResults for pages and containers.
	 *
	 * @param Hits hits, the search result returned by lucene.
	 * @param ParamBean param, the param bean used to control access.
	 * @param int results the param bean.
	 * @return JahiaSearchResult result, the result.
	 */
	private JahiaSearchResult buildContainerSearchResult( Hits hits,
													  	  ParamBean jParams ) {

        JahiaSearchResult result = new JahiaSearchResult();

		if ( hits == null || hits.length()==0 )
			return result;

		ServicesRegistry sReg = ServicesRegistry.getInstance();
		JahiaField aField = null;
		Object obj = null;
		int objID = 0;
		JahiaUser currentUser = jParams.getUser();
        JahiaSearchHit      info;

		int nbHits = hits.length();
		String fieldIDVal = null;
		boolean found = false;

		for (int i=0; i<nbHits ; i++) {
			try {
				fieldIDVal = hits.doc(i).get(JahiaSearchConstant.FIELD_FIELDID);
				if ( fieldIDVal != null ){
					aField = sReg.getJahiaFieldService()
							 .loadField(Integer.parseInt(fieldIDVal), jParams);
			  	}
			} catch (Throwable t) {
				//
			}

			// security check over the field.
			if ( (aField != null) && aField.checkReadAccess(currentUser,jParams.getSiteID()) )
			{
		  		found = false;
		  		int size = result.results().size();
				for (int j=0; j<size; j++) // page already found
				{
				   if (((JahiaSearchHit)result.results().elementAt(j)).id == aField.getctnid())
				   {
				       ((JahiaSearchHit)result.results().elementAt(j)).wordcount += 1;
				       found = true;
				       break;
				   }
				}

				objID = aField.getctnid();

				if (!found)
				{
				   info = new JahiaSearchHit();
				   info.wordcount = 1;

				   info.id = objID;
				   info.obj = new Integer(objID); // actually we do not load container

				   String value = aField.getValue();
				   if (value!=null){
				       try {
				           value = (new RE("<(.*?)>")).subst(value,"");
				       } catch (RESyntaxException e) {
				           JahiaConsole.println(CLASS_NAME+".buildContainerSearchResult",e.toString());
				       } catch (Throwable t) {
				           JahiaConsole.println(CLASS_NAME+".buildContainerSearchResult",t.toString());
				       }
				       info.teaser = GuiBean.glueTitle(value,JahiaSearchConstant.TEASER_LENGTH);

				   } else {
				       info.teaser = "";
				   }
				   result.addHit(info);
				}
			}
			fieldIDVal = null;
			aField = null;
			obj = null;
		}

		// sort the result
		if ( result.results().size()>1 ){
        	// a dummy hit
        	JahiaSearchHit dummyHit = new JahiaSearchHit();
        	dummyHit.wordcount = 0;
        	Collections.sort(result.results(),dummyHit);
		}

		return result;
	}

	//--------------------------------------------------------------------------
	/**
	 * Build jahia search result ( matching pages ) from hits returned by lucene.
	 *
	 * TODO : have a common method for building JahiaSearchResults for pages and containers.
	 *
	 * @param Hits hits, the search result returned by lucene.
	 * @param ParamBean param, the param bean used to control access.
	 * @return JahiaSearchResult result, the result.
	 */
	private JahiaSearchResult buildJahiaSearchResult( Hits hits,
													  ParamBean jParams ){

        JahiaSearchResult result = new JahiaSearchResult();

		if ( hits == null || hits.length()==0 )
			return result;

		ServicesRegistry sReg = ServicesRegistry.getInstance();
		JahiaField aField = null;
		JahiaPage aPage = null;
		JahiaUser currentUser = jParams.getUser();
        JahiaSearchHit      info;

		int nbHits = hits.length();
		String fieldIDVal = null;
		boolean found = false;
		int pagecount = 0;

		for (int i=0; i<nbHits ; i++) {
			try {
				fieldIDVal = hits.doc(i).get(JahiaSearchConstant.FIELD_FIELDID);
				if ( fieldIDVal != null ){
					aField = sReg.getJahiaFieldService()
							 .loadField(Integer.parseInt(fieldIDVal), jParams);
			      	aPage = sReg.getJahiaPageService().lookupPage (aField.getPageID(), jParams);
			  	}
			} catch (Throwable t) {
				//
			}

			// security check over the field and the corresponding page
			if ( aPage != null && aPage.checkReadAccess(currentUser) && aField.checkReadAccess(currentUser,jParams.getSiteID()) )
			{
		  		found = false;
				for (int j=0; j<result.getHitCount(); j++) // page already found
				{
				   if (((JahiaSearchHit)result.results().elementAt(j)).id == aField.getPageID())
				   {
				       ((JahiaSearchHit)result.results().elementAt(j)).wordcount += 1;
				       found = true;
				       break;
				   }
				}
				if (!found)
				{
				   info = new JahiaSearchHit();
				   info.wordcount = 1;

				   info.id = aPage.getID();
				   info.obj = aPage;

				   //Keep for backward compatibility ---------------------------
				   info.page = aPage;
				   info.pageID = aPage.getID();
				   //-----------------------------------------------------------

				   String value = aField.getValue();
				   if (value!=null){
				       try {
				           value = (new RE("<(.*?)>")).subst(value,"");
				       } catch (RESyntaxException e) {
				           JahiaConsole.println(CLASS_NAME+".buildJahiaSearchResult",e.toString());
				       } catch (Throwable t) {
				           JahiaConsole.println(CLASS_NAME+".buildJahiaSearchResult",t.toString());
				       }
				       info.teaser = GuiBean.glueTitle(value,JahiaSearchConstant.TEASER_LENGTH);

				   } else {
				       info.teaser = "";
				   }
				   result.addHit(info);
				   pagecount++;
				}
			}
			fieldIDVal = null;
			aField = null;
			aPage = null;
		}

		// sort the result
		if ( result.getHitCount()>1 ){
        	// a dummy hit
        	JahiaSearchHit dummyHit = new JahiaSearchHit();
        	dummyHit.wordcount = 0;
        	Collections.sort(result.results(),dummyHit);
		}

		//Keep for backward compatibility ---------------------------
        result.pagecount = pagecount;
		result.pages = result.results();
		//-----------------------------------------------------------

		return result;
	}

}