sitefilter.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 Dec 1, 2005
 * Contact: [e-mail]
 * *****************************************************
 */
package org.indigo.db.fieldfilter;

/**
 * 处理URL主要是获取一个URL中的站点域名。
 * @author wbz
 *
 */
public class SiteFilter
{
    private static SiteFilter itsInstance=new SiteFilter();
    private SiteFilter()
    {
        
    }
    public static SiteFilter getInstance()
    {
        return itsInstance;
    }
    public String filter( String site )
    {
        String prefix="http://",postfix="/";
        site = site.toLowerCase();
        int i,j;
        i = site.indexOf( prefix );
        if( i==-1 )
        {
            i = 0;
            j = site.indexOf( postfix );
        }else
        {
            i += prefix.length();
            j = site.indexOf( postfix, i+1 );
        }
        if( i!=-1 && j!=-1 && i<=j )
            site = site.substring( i, j );
        
        return site;
    }
}