filtuntil.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.filters;
/**
 * 实现了Filter接口，过滤掉html中的<>标签。
 * @author wbz
 *
 */
public class FiltUntil implements Filter
{
    private String itsStr;
    public FiltUntil( String str )
    {
        itsStr = str;
    }
    public String filt( String str )
    {
        int i1=-1, i2=-1;
        i1=str.indexOf("<");
        i2 = str.indexOf( itsStr );
        if( i2==-1 )
            return str;
        if( i1<i2 && i1!=-1 && i2!=-1  )
            return str;
        str = str.substring( i2+1 );
        return str;
    }

}