VisClient/org/hfbk/vis/source/SourceGoogle.java

Go to the documentation of this file.

package org.hfbk.vis.source;

import java.net.URL;

import org.dronus.graph.Node;
import org.hfbk.util.HTMLUtils;
import org.hfbk.util.HTTPUtils;

public class SourceGoogle extends SourceRegExp {
        
        public SourceGoogle(URL url) {          
                super(url);
                root=add(keyword,"keyword",null);               
        }

        String url(){
                return "http://www.google.com/search?q="+HTTPUtils.encode(keyword);
        }
        
        String matcher(){
                return "(?si)class=r><a href=\"(.*?)\".*?>(.*?)</a>.*?<div class=(.*?)>(.*?)<br";
        }
        
        void parse(String[] matches){
                String text=HTMLUtils.clean(matches[4]);
                
                if (text.length()==0) return;
                
                Node textNode=add(text,"text",root);
                add(matches[1], "URL" ,textNode);
                add(HTMLUtils.clean(matches[2]), "headline", textNode); 
                
                //split out keywords
                for(String kw : text.split(" ")){
                        //$kw=preg_replace('§[^\wüäöÜÄÖß]§','',$kw);
                        if (kw.length()>3) 
                                add(HTMLUtils.removePunctuation(kw),"keyword",textNode);
                }
        }       
}

---