articleclassify.java

用java实现的一个bbs的portal

/**
 * Created by IntelliJ IDEA.
 * User: Administrator
 * Date: Jul 17, 2003
 * Time: 11:08:34 AM
 * To change this template use Options | File Templates.
 */
package Classification;

import Kernel.Article;
import Kernel.TradeInfo;

import java.util.Vector;
import java.util.Enumeration;
import java.util.StringTokenizer;

public class ArticleClassify {

    private Classifier classifier;
    private String FileName;

    public ArticleClassify(String filename){
        this.classifier = new Classifier();
        this.FileName = filename;
    }

    public boolean DoClassify(){
        ArticleDomParser domParser = new ArticleDomParser(FileName);
        if (!domParser.isContinue()) {
            System.out.println(ArticleBatch.moveErrorFile(FileName));
            System.err.println("File "+FileName+" Dom Parse Error and Move into ErrorArticle .");
            return false;
        }
        Vector ArticleVec = domParser.getArticleVec();
        Article article = (Article)ArticleVec.get(0);
        return storeInfo(article);
    }

    private short getTradeType(Article article){   //0--sale,1--buy,2--lease,3--wantlease
        String title = article.getTitle();
        String result = classifier.checkTradeType(title);
        if (result.equals("SALE"))
            return 0;
        else if  (result.equals("BUY"))
            return 1;
        else if (result.equals("LEASE"))
            return 2;
        else if (result.equals("WANTLEASE"))
            return 3;
        else return -1;
    }

    private Vector getCommodity(Article article){
        String title = article.getTitle();
        String content = article.getContent();
        Vector result1 = classifier.checkTitleCommodity(title);
        if (result1.size() == 0){
            Vector result2 = classifier.checkContentCommodity(content);
            return result2;
        }else {
            classifier.checkContentContact(content);
            return result1;
        }
    }

    private boolean storeInfo(Article article){
        boolean isSuccess = true;
        //交易信息分类
        short TradeType = getTradeType(article);
        /*
        if (TradeType == -1)
        return false;
        */
        Vector Commodity = getCommodity(article);
        Vector Contact = classifier.contactVec;
        String contactStr = " ";
        for (Enumeration e = Contact.elements(); e.hasMoreElements();) {
            contactStr = contactStr + (String)e.nextElement() + " ";
        }

        if (Commodity.size() != 0){
            boolean HouseOnly = false;
            for (Enumeration e = Commodity.elements();e.hasMoreElements();){
                String commodityItem =  (String)e.nextElement();
                StringTokenizer st = new  StringTokenizer(commodityItem,"#");
                String commodity = st.nextToken();
                String description = st.nextToken();
                for (Enumeration ee = ClassifyConfig.getInstance().HouseCommodityWords.elements(); ee.hasMoreElements();) {
                    if (commodity.equalsIgnoreCase((String)ee.nextElement())){
                        HouseOnly = true;
                        break;
                    }
                }
                if (HouseOnly){
                    boolean notHouse = true;
                    for (Enumeration ee = ClassifyConfig.getInstance().HouseCommodityWords.elements(); ee.hasMoreElements();) {
                        if (commodity.equalsIgnoreCase((String)ee.nextElement())){
                            notHouse = false;
                            break;
                        }
                    }
                    if (notHouse)
                        continue;
                }
                for (Enumeration ee = ClassifyConfig.getInstance().OnlyLeaseWords.elements();ee.hasMoreElements();){
                    String leaseWord = (String)ee.nextElement();
                    if (commodity.indexOf(leaseWord) != -1){
                        if (TradeType == 0)
                            TradeType =2;
                        else if (TradeType ==1)
                            TradeType =3;
                    }
                }
                TradeInfo tradeinfo = new TradeInfo(TradeType,commodity,
                        description,contactStr,article.getDate(),article.getXMLFile());
                if (tradeinfo.insertTradeInfo())
                    System.out.println("Trade Info "+article.getXMLFile()+" Info Insert successfully !");
                else {
                    System.err.println("Trade Info "+article.getXMLFile()+" Info Insert Failed !");
                    isSuccess = false;
                }
            }
        }else {
            String commodity = "NULL";
            String description = article.getTitle()+"<$TITLE$>"+article.getContent();
            TradeInfo tradeinfo = new TradeInfo(TradeType,commodity,
                    description,contactStr,article.getDate(),article.getXMLFile());
            if (tradeinfo.insertTradeInfo())
                System.out.println("Trade Info "+article.getXMLFile()+" Info Insert successfully !");
            else {
                System.err.println("Trade Info "+article.getXMLFile()+" Info Insert Failed !");
                isSuccess = false;
            }
        }
        if (!isSuccess){
            TradeInfo.delTradeInfo(article.getXMLFile());
        }else{
            //将BBS原文插入数据库
            if (article.insertBBSInfo())
                System.out.println("Article "+article.getXMLFile()+" Info Insert successfully !");
            else System.err.println("Article "+article.getXMLFile()+" Info Insert Failed !");
        }
        return isSuccess;
    }

    public static void main(String args[]){
        System.out.println((new ArticleClassify("bbs.pku.edu.cn_SecondHand_2003-09-14 19-43-19.xml")).DoClassify());
    }

}

/*
private Vector getTitleFromFile(String filepath)
{
Vector vec=new Vector();
try{
BufferedReader reader=new BufferedReader(new FileReader(filepath));
String str=null;
while ((str=reader.readLine())!=null){
vec.add(str);
}
reader.close();
}
catch (Exception e) {
System.err.println("ERROR:Can not read file "+filepath+" !");
e.printStackTrace();
return null;
}
return vec;
}
*/

/*
public Vector getCommodity(){
String title = article.getTitle();
//String content = article.getContent();
TestSlicer tSlicer=new TestSlicer();  //直接读入一个字符串
Vector TitleWords = tSlicer.doSlicer(title);
Vector kinds = new Vector();
int wordNum = 0;
int goodsNum = -1;
for (Enumeration e = TitleWords.elements();e.hasMoreElements();){
Word word = (Word)e.nextElement();
wordNum++;
if (word.getAttri().equalsIgnoreCase("n")) {
String result = ClassifyConfig.getInstance().checkCommodity(word.getContent());
if (result != null){
if (goodsNum == wordNum-1){
kinds.set(kinds.size()-1,result);
}else{
kinds.add(result);
}
goodsNum = wordNum;
}
}
if (word.getAttri().equalsIgnoreCase("u")){
if (kinds.size()>0)
kinds.remove(kinds.size()-1);
}
}
return kinds;
}
*/