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

Go to the documentation of this file.

package org.hfbk.vis.source;

import java.io.IOException;
import java.net.URL;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

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

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

        String url(){
                return "http://www.google.de/search?btnI=yes&q=inurl:wikipedia.org+"+HTTPUtils.encode(keyword);
        }
        
        String matcher(){
                return "(?si)(?:href=\"/wiki/([^:\"]*?)\")|(?:src=\"(http://upload.[^\"]*?)\")";
        }
        
        void buildGraph() throws IOException {          
                
                String page=HTTPUtils.fetch(url(), silent);
                
                page=page.substring(page.indexOf("<!-- start content -->"), page.lastIndexOf("<!-- end content -->"));
        
                Pattern p=Pattern.compile(matcher());
                Matcher matcher=p.matcher(page);
                while (matcher.find()){
                        String imgMatch=matcher.group(2);
                        if(imgMatch!=null) {
                                String imgUrl=imgMatch.replace("thumb/","");
                                imgUrl=imgUrl.substring(0,imgUrl.lastIndexOf('/'));
                                Node img=add(imgUrl,"image",root);
                                add(imgMatch, "thumbnail", img);
                        }
                        else add(matcher.group(1),"keyword",root);
                }
        }
        
        //unused
        void parse(String[] matches) throws IOException {}
}

---