testfilters.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.filters;

import org.indigo.filters.FiltEnter;
import org.indigo.filters.FiltFrontBack;
import org.indigo.filters.FiltSpace;
import org.indigo.filters.FiltUntil;
import org.indigo.filters.Filter;

import junit.framework.TestCase;

public class TestFilters extends TestCase
{
    private Filter [] filters = null;
    
    public String filt( String str )
    {
/*        
        filters = new Filter[4];
        filters[0] = new FiltUntil( ">" );
        filters[1] = new FiltFrontBack();
        filters[2] = new FiltSpace();
        filters[3] = new FiltEnter();
        
        int tmpsz,j;
        tmpsz = filters.length;
        for( j=0; j<tmpsz; j++ )
            str = filters[j].filt( str );
*/
        Filter theFlt = null;
        theFlt = new FiltUntil( ">" );
        str = theFlt.filt( str );

        theFlt = new FiltFrontBack();
        str = theFlt.filt( str );

        theFlt = new FiltSpace();
        str = theFlt.filt( str );

        theFlt = new FiltEnter();
        str = theFlt.filt( str );

        return str;
    }
    
    public void testFilters()
    {
        String str=null;
        str = "</font></td><td width=\"246\" height=\"22\" bgcolor=\"#FFFFFF\"><font  size=\"2\">&nbsp;岚县县城内西三街</font></td></tr><tr><td width=\"60\" height=\"22\" bgcolor=\"#E6E6E6\" align=\"center\"><font size=\"2\">";
        
        str = filt( str );
        System.out.println( str );
        assertTrue( str.equals("岚县县城内西三街") );
    }
}