testparser.java

用来为垂直搜索引擎抓取数据的采集系统

/*
 * *****************************************************
 * Copyright (c) 2005 IIM Lab. All  Rights Reserved.
 * Created by xuehao at 2005-10-12
 * Contact: [e-mail]
 * *****************************************************
 */

package org.indigo.tests.parser;

import org.indigo.pages.VisitPage;
import org.indigo.parser.Parser;

import junit.framework.TestCase;

public class TestParser extends TestCase
{
    public void testParser()
    {
        Parser parser = new Parser();
        
        parser.setUrl( "http://www.ahnw.gov.cn/scxx/schq/?datetime=&page=2&zl=&diqu=&chanpin=&dl=&NewDay=0" );
        parser.open();
        
        String midStr=null,startStr=null, endStr=null;
/*        
        startStr = "<td class=\"z\" width=\"24%\" height=20 style=\"border-right:1 solid #FFFFFF;border-bottom: 1 solid #FFFFFF\">&nbsp;";
        endStr = "</td>";
*/
        startStr="<td width=\"11%\" class=\"z\" style=\"border-right:1 solid #FFFFFF;border-bottom: 1 solid #FFFFFF\">&nbsp;";
        endStr="</td>";
        
        midStr = parser.parseWith( startStr, endStr );
        System.out.println( midStr );
        assertTrue( midStr.equalsIgnoreCase("元/公斤") );
        
        parser.close();
        
    }

}