topicimpl.java

这是一款最新的野蔷薇论坛源码,有需要的朋友可以尽情下载

/* 
 * Created on 2007-5-27
 * Last modified on 2007-6-10
 * Powered by GamVan.com
 */
package com.yeqiangwei.club.service.search.impl;

import java.io.IOException;
import java.io.StringReader;
import java.util.ArrayList;
import java.util.List;

import org.apache.log4j.Logger;
import org.apache.lucene.analysis.TokenStream;
import org.apache.lucene.document.Document;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.queryParser.ParseException;
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.apache.lucene.search.BooleanClause;
import org.apache.lucene.search.BooleanClause.Occur;
import org.apache.lucene.search.highlight.Fragmenter;
import org.apache.lucene.search.highlight.Highlighter;
import org.apache.lucene.search.highlight.QueryScorer;
import org.apache.lucene.search.highlight.SimpleFragmenter;
import org.apache.lucene.search.highlight.SimpleHTMLFormatter;

import com.yeqiangwei.club.param.TopicParameter;
import com.yeqiangwei.club.service.ServiceLocator;
import com.yeqiangwei.club.service.ServiceWrapper;
import com.yeqiangwei.club.service.model.ContentModel;
import com.yeqiangwei.club.service.model.TopicModel;
import com.yeqiangwei.club.service.search.AnalyzerFactory;
import com.yeqiangwei.club.service.search.Constants;
import com.yeqiangwei.club.service.search.MultiFiledQuery;
import com.yeqiangwei.club.service.search.SearchParameter;
import com.yeqiangwei.club.service.search.SearchProvider;
import com.yeqiangwei.club.service.search.SearchUtils;
import com.yeqiangwei.club.service.topic.TopicService;
import com.yeqiangwei.club.service.util.BasicInfoService;
import com.yeqiangwei.io.File;
import com.yeqiangwei.util.StringHelper;
import com.yeqiangwei.util.TypeChange;
import com.yeqiangwei.util.Validator;


@SuppressWarnings("unchecked")
public class TopicImpl extends SearchProvider{
	
	private static final Logger logger = Logger.getLogger(TopicImpl.class);
	
	private static int CANSEARCH = 1;
	
	private static String MINID_FILE = null;
	
	private static String INDEXPATH_TOPIC = null;
	
	private static String TEMPPATH_TOPIC = null;
	
	static
	{
		INDEXPATH_TOPIC = getBasicService().findOnly().getIndexPath()+"topics";
		TEMPPATH_TOPIC = Constants.INDEX_PATH+"topics"+File.separator;
		MINID_FILE = Constants.INDEX_PATH+"topics.txt";
	}
	
	public static void main(String args[]){
		com.yeqiangwei.club.dao.hibernate.ConnectionManager.init();
		com.yeqiangwei.club.util.LanguageUtils.instance();
		TopicImpl st = new  TopicImpl();
		
		st.createIndex(1);
		System.out.println("Job over");
		
		/*
		List<TopicModel> list = st.results(1,30,"美女 喜欢");
		for(int i=0; i<list.size(); i++){
			TopicModel model = list.get(i);
			System.out.println(model.getTitle());
			System.out.println(model.getContentModel().getContent());
		}*/
		System.out.println("========================");
	}
	
	private TopicModel filter(TopicModel model){
		com.yeqiangwei.club.dao.hibernate.ConnectionManager.beginTransaction();
		/*
		 * 如果标题表topicId和内容表的不符则依据标题表topicId去查找内容
		 * 
		 */
		int topicId = model.getTopicId();
		ContentModel content = model.getContentModel();
		if(Validator.isEmpty(content)){
			model = this.getTopicService().findTopicAndContentById(topicId);
		}else{
			int ctopicId = content.getTopicId();
			if(topicId!=ctopicId){
				model = this.getTopicService().findTopicAndContentById(topicId);
				if(this.getTopicService().findById(ctopicId)==null){ //找不到标题的内容
					int c = this.getTopicService().getContentDAO().deleteByTopicId(ctopicId);
					logger.info("删除找不到标题的内容 topicId="+ctopicId);
					if(c>0){
						logger.info("删除成功！");
					}else{
						logger.info("删除失败！");
					}
				}
			}
		}
		com.yeqiangwei.club.dao.hibernate.ConnectionManager.commitTransaction();
		return model;
	}
	
	private void addDocs(TopicModel model, IndexWriter writer) throws IOException
	{
		model = this.filter(model);
		if(Validator.isEmpty(model)||Validator.isEmpty(model.getTitle())){
			logger.warn("找不到帖子！");
		}else{
			/*
			System.out.println(model.getTopicId()+":"+model.getContentModel().getTopicId());
			System.out.println("======================================");
			System.out.println("");
			*/
			Constants.TOPICID = model.getTopicId();
			Document doc = new Document();
			doc.add(SearchUtils.Text("topicId",model.getTopicId()));
			doc.add(SearchUtils.Text("userId",model.getUserId()));
			doc.add(SearchUtils.Text("userName",model.getUserName()));
			doc.add(SearchUtils.Text("createDateTime",model.getCreateDateTime()));
			doc.add(SearchUtils.Text("title",model.getTitle()));
			doc.add(SearchUtils.Text("content",model.getContentModel().getContent()));
			doc.add(SearchUtils.Text("forumId",model.getForumId()));
			writer.addDocument(doc);
			File.createFile(MINID_FILE,String.valueOf(model.getTopicId()),"gb2312",true);
		}
	}
	

	public void createIndex(int auto) {
		if(state==1){
			logger.info("任务重复！");
			return ;
		}
		state=1;
		if(auto==1){ //自动判断是创建还是追加
			int id = 0;
			try {
				String strId = File.readTxt(MINID_FILE,"gb2312");
				logger.info("MINID_FILE: "+strId);
				id = TypeChange.stringToInt(strId);
			} catch (IOException e) {
				logger.error(e.toString());
			}
			if(id>0){
				createIndex(false,id);
			}else{
				createIndex(true,id);
			}
		}else{ //强制重新生成索引
			try {
				File.createFile(MINID_FILE,"0","gb2312",true);
				createIndex(true,0);
			} catch (IOException e) {
				logger.error(e.toString());
			}
		}
		state=2;
	}


	public void createIndex(boolean iscreate, int minId) {
		try {
			logger.info("索引创建开始！");
			logger.debug(TEMPPATH_TOPIC);
			int maxBufferedDocs = getBasicService().findOnly().getMaxBufferedDocs();
			int mergeFactor = getBasicService().findOnly().getMergeFactor();
			int maxFieldLength = getBasicService().findOnly().getMaxFieldLength();
			logger.debug("maxBufferedDocs: "+maxBufferedDocs);
			logger.debug("mergeFactor: "+mergeFactor);
			logger.debug("maxFieldLength: "+maxFieldLength);
			File.createFolder(TEMPPATH_TOPIC);
			IndexWriter writer = SearchUtils.getIndexWriter(TEMPPATH_TOPIC, iscreate);
			writer.setMaxBufferedDocs(maxBufferedDocs); // Defalut 10 
			writer.setMergeFactor(mergeFactor); // Defalut 10 
			writer.setMaxFieldLength(maxFieldLength);
			int rows = 200;
			TopicParameter param = new TopicParameter();
			param.setIsDeleted(false);
			param.setOrderBy(new Byte((byte) 4));
			param.setRows(new Integer(rows));
			param.setMinId(new Integer(minId));
			logger.info("rows: "+rows);
			logger.info("minId: "+minId);
			long count = this.getTopicService().countByParameter(param);
			int pages = (int) (count/rows+1);
			logger.info("pages: "+pages);
			outer: 
			for(int page=1; page<=pages; page++){
				param.setPage(new Integer(page));
				List<TopicModel> list = this.getTopicService().findByParameter(param);
				List<ContentModel> clist = this.getTopicService().findContentByParameter(param);
				if(!Validator.isEmpty(list)&&!Validator.isEmpty(clist)){
					if(list.size()==clist.size()){
						for(int i=0; i<list.size(); i++){
							TopicModel model = list.get(i);
							ContentModel cmodel = clist.get(i);
							model.setContentModel(cmodel);
							this.addDocs(model,writer);
							if(Constants.STOP==1){
								state = 0;
								break outer;
							}
						}
					}else{
						logger.warn("==============================");
						logger.warn("取出的文章标题数和文章内容不符");
						logger.warn("文章数："+list.size());
						logger.warn("内容数："+clist.size());
						logger.warn("当前页："+page);
						logger.warn("==============================");
						for(int i=0; i<list.size(); i++){
							TopicModel model = list.get(i);
							this.addDocs(model,writer);
							if(Constants.STOP==1){
								state = 0;
								break outer;
							}
						}
					}
				}
				if(Constants.STOP==1){
					state = 0;
					break outer;
				}
			}
			writer.optimize();
			writer.close();
			if(Constants.STOP!=1){
				logger.info("终止搜索服务！");
				CANSEARCH = 0;
				logger.info("删除目录旧索引！"+INDEXPATH_TOPIC);
				File.deleteFolder(INDEXPATH_TOPIC);
				logger.info("移动新索引到目录："+INDEXPATH_TOPIC);
				File.copyFolder(TEMPPATH_TOPIC,INDEXPATH_TOPIC);
				logger.info("开启索引服务器！");
				CANSEARCH = 1;
				logger.info("索引创建完成，本次任务结束！");
				state=2;
			}else{
				logger.info("索引创建任务被强迫终止，本次任务结束！");
			}
		} catch (IOException e) {
			logger.error(e.toString());
		}
	}

	public void deleteIndex(int id) {
		CANSEARCH = 0;
		SearchUtils.delete(id,"topicId",INDEXPATH_TOPIC);
		SearchUtils.delete(id,"topicId",TEMPPATH_TOPIC);
		CANSEARCH = 1;
	}

	public List<TopicModel> results(SearchParameter param) {
		if(CANSEARCH==0){
			return null;
		}
		super.totalrows = 0;
		super.runtimed = 0;
		long begin = System.currentTimeMillis();
		List<TopicModel> list = null;
		try {
			List<String> fields = new ArrayList<String>();
			List<BooleanClause.Occur> flags = new ArrayList<BooleanClause.Occur>();
			List<Float> boosts = new ArrayList<Float>();
			List<String> keys = new ArrayList<String>();
			fields.add("title");
			fields.add("content");
			fields.add("userName");
			boosts.add(new Float(0.9));
			boosts.add(new Float(0.5));
			boosts.add(new Float(0.1));
			keys.add(param.getKeys());
			keys.add(param.getKeys());
			keys.add(param.getKeys());
			Searcher searcher = new IndexSearcher(INDEXPATH_TOPIC);
			list = new ArrayList<TopicModel>();
			if(!Validator.isEmpty(param.getForumId())&&param.getForumId().intValue()>0){
				fields.add("forumId");
				boosts.add(new Float(1));
				keys.add(String.valueOf(param.getForumId()));
				flags.add(Occur.MUST);
				flags.add(Occur.SHOULD);
				flags.add(Occur.SHOULD);
				flags.add(Occur.MUST);
				
			}else{
				flags.add(Occur.SHOULD);
				flags.add(Occur.SHOULD);
				flags.add(Occur.SHOULD);
			}
			Query query = MultiFiledQuery.parse(keys, fields, flags, boosts, AnalyzerFactory.getAnalyzer());
			Hits hits = searcher.search(query);
			SimpleHTMLFormatter sf = new SimpleHTMLFormatter("<font class=\"red\">","</font>");
			QueryScorer scorer = new QueryScorer(query);
			Highlighter highlighter_title = new Highlighter(sf, scorer);
			Fragmenter fragmenter = new SimpleFragmenter(100);
			highlighter_title.setTextFragmenter(fragmenter);
			Highlighter highlighter_content = new Highlighter(sf, scorer);
			highlighter_content.setTextFragmenter(new SimpleFragmenter(5));
			if(hits!=null){
				super.totalrows = (hits.length());
				int listrows = param.getRows().intValue(); //每页显示行数
				int startrows = 0; //显示记录集的起始行
				int maxrows = listrows; //当前页最大行数
				/* 翻页滤错开始 */
				if(listrows>totalrows){
					listrows = totalrows;
				}
				if(param.getPage().intValue()>1){
					startrows = (listrows * (param.getPage().intValue()-1));
					maxrows = startrows + listrows;
				}
				if(maxrows > totalrows){
					maxrows = totalrows;
				}
				for (int i = startrows; i < maxrows; i++) {
					try {
						Document doc = hits.doc(i);
						String title = doc.get("title");
						title = StringHelper.htmlEncoder(title);
						TokenStream stream = AnalyzerFactory.getAnalyzer().tokenStream("title",new StringReader(title));
						title = highlighter_title.getBestFragment(stream, title);
						String content = doc.get("content");
						content = StringHelper.htmlEncoder(content);
						stream = AnalyzerFactory.getAnalyzer().tokenStream("content",new StringReader(content));
						content = highlighter_content.getBestFragments(stream,content,50,"...");
						content = com.yeqiangwei.util.StringHelper.ubbPattern(content,"\\[(.*?)\\](.*?)\\[/(.*?)\\]","$2","");
						if(Validator.isEmpty(title)){
							title = doc.get("title");
							title = StringHelper.htmlEncoder(title);
						}
						if(Validator.isEmpty(content)){
							content = StringHelper.substring(doc.get("content"),0,50,"");
							content = StringHelper.htmlEncoder(content);
						}/*else{
							content = content.replaceAll("<br/>"," ");
							if(content.length()>200){
								content = StringHelper.substring(content,0,200,"")+"...";
							}
						}*/
						content = content.replaceAll("<br/>","...");
						content = content.replaceAll("&nbsp;"," ");
						StringHelper.ubbPattern(content,"(\\s+)"," ","");
						TopicModel tmodel = new TopicModel();
						tmodel.setTopicId(TypeChange.stringToInt(doc.get("topicId")));
						tmodel.setTitle(title);
						tmodel.setUserId(TypeChange.stringToInt(doc.get("userId")));
						tmodel.setUserName(doc.get("userName"));
						tmodel.setForumId(TypeChange.stringToInt(doc.get("forumId")));
						ContentModel cmodel = new ContentModel();
						cmodel.setContent(content);
						tmodel.setContentModel(cmodel);
						list.add(tmodel);
					} catch (IOException e) {
						logger.error(e.toString());
					}  
				}
			}
			searcher.close();
		} catch (IOException e) {
			logger.error(e.toString());
		} catch (ParseException e) {
			logger.error(e.toString());
		}
		long end = System.currentTimeMillis();
		super.runtimed = (end - begin);
		return list;
	}

	public TopicService getTopicService() {
		return ServiceWrapper.<TopicService>getSingletonInstance(ServiceLocator.TOPIC);
	}
	
	public static BasicInfoService getBasicService() {
		return ServiceWrapper.<BasicInfoService>getSingletonInstance(ServiceLocator.BASICINFO);
	}
	
	public int getState() {
		return state;
	}

	public long getRuntimed() {
		return runtimed;
	}
	
	public int getTotalrows() {
		return super.totalrows;
	}

	public void setTotalrows(int totalrows) {
		super.totalrows = totalrows;
	}
}