source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/GoogleSearch.java.tmp@ 11228

package org.greenstone.gsdl3.service;

// Greenstone classes
import org.greenstone.gsdl3.util.*;

//Google Web Services API classes
//import com.google.soap.search.GoogleSearch;
import com.google.soap.search.GoogleSearchFault;
import com.google.soap.search.GoogleSearchResult;
import com.google.soap.search.GoogleSearchResultElement;
import com.google.soap.search.GoogleSearchDirectoryCategory;

// XML classes
import org.w3c.dom.Element; 
import org.w3c.dom.Document;
import org.w3c.dom.NodeList; 

//Java classes
import java.util.ArrayList;
import java.util.HashMap;
import java.io.File;
import java.io.InputStream;
import java.io.BufferedReader;
import java.io.InputStreamReader;
import java.io.IOException;
import java.net.HttpURLConnection;
import java.net.URLConnection;
import java.net.URL;
import java.net.Authenticator;
import java.net.PasswordAuthentication;
import java.net.MalformedURLException;
import java.lang.Object;


/**
 *
 * @author <a href="mailto:[email protected]">Chi-Yu Huang</a>
 *
 */

public class GoogleSearch
    extends AbstractSearch {
    
    //Parameters connect to Proxy Server
    private boolean using_proxy = false;
    private String proxy_host = null;
    private int proxy_port;
    private String proxy_user = null;
    private char [] proxy_passwd = null;

    // google key
    private String client_key = null;
    
    public GoogleSearch()
    {
    }
        
    //Configure GoogleSearch Service
    public boolean configure(Element info, Element extra_info) 
    { 
    if (!super.configure(info, extra_info)) {
        return false;
    }
    System.err.println("Configuring GoogleSearch");
    Element server_elem = (Element)GSXML.getChildByTagName(info, "googleServer");
    if (server_elem == null) {
        System.err.println("GoogleSearch.configure error: no googleServer element found");
        return false;
    }
    
    client_key = server_elem.getAttribute("key");
    if (client_key.equals("")) {
        System.err.println("GoogleSearch.configure error: no client_key for the googleServer element");
        return false;
    }
    
    does_paging = true;
    does_chunking = false;

    // are we behind a proxy??
    // all the details should have been set up by the Message Router
    proxy_host = System.getProperty("http.proxyHost");
    if (proxy_host != null && !proxy_host.equals("")) {
        using_proxy = true;
        try {
        proxy_port = Integer.parseInt(System.getProperty("http.proxyPort").trim());
        } catch (Exception e) {
        System.err.println("GoogleSearch.configure error: couldn't get proxy port, defaulting to 80");
        proxy_port = 80;
        }
        PasswordAuthentication pa = Authenticator.requestPasswordAuthentication(proxy_host, null, proxy_port, "http", "", null);
        proxy_user = pa.getUserName();
        proxy_passwd = pa.getPassword();
    }
    return true;
    }
    
    /** Process a text query - implemented by concrete subclasses */
    protected Element processTextQuery(Element request) {
    
    //Connect to Google Web API service
    com.google.soap.search.GoogleSearch search = new com.google.soap.search.GoogleSearch();

    // Set mandatory attributes
    search.setKey(client_key);

    // proxy??
    if (using_proxy) {
        search.setProxyHost(proxy_host);
        search.setProxyPort(proxy_port);
        search.setProxyUserName(proxy_user);
        search.setProxyPassword(new String(proxy_passwd));
    }

    //set optional attributes
    search.setSafeSearch(true);
    
    // Create a new (empty) result message
    Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
    result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
    result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
    Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
    result.appendChild(doc_node_list);
    Element query_metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
    result.appendChild(query_metadata_list);

    Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
    if (param_list == null) {
        System.err.println("GoogleSearch.processTextQuery Error:: TextQuery request had no paramList.");
        return result;  // Return the empty result
    }

    // Process the request parameters
    HashMap params = GSXML.extractParams(param_list, false);

    // Make sure a query has been specified
    String query = (String) params.get(QUERY_PARAM);
    if (query == null || query.equals("")) {
        System.err.println("GoogleSearch.processTextQuery Error: TextQuery request had no query string.");
        return result;  // Return the empty result
    }
        // tidy whitespace
    query = query.replaceAll("\\s+", "+");
    
    search.setQueryString(query);

    //Check hits_per_page
    int hits_per_page;
    try {
        hits_per_page = Integer.parseInt(((String)params.get(HITS_PER_PAGE_PARAM)).trim());
    } catch (Exception e) {
        System.err.println("GoogleSearch.processTextQuery error: couldn't get hits per page param, defaulting to 10");
        hits_per_page = 10;
    }
    //Check the start_page number
    int start_page;
    try {
        start_page = Integer.parseInt(((String) params.get(START_PAGE_PARAM)).trim());
    } catch (Exception e) {
        System.err.println("GoogleSearch.processTextQuery error: couldn't get start page param, defaulting to 1");
        start_page = 1;
    }
    
    //Invoke Actual Search
    
    // Google only allows 10 hits per request
    int loop = hits_per_page/10;
    int remainder = hits_per_page%10;
    int google_start_page = (start_page-1)*hits_per_page;
    int pages_per_loop;
    for (int j=0; j < loop; j++){
        if (j < (loop-1) || remainder == 0) {
        pages_per_loop = 10;
        } else {
        pages_per_loop = remainder;
        }
        search.setMaxResults(pages_per_loop); 
        search.setStartResult(google_start_page);
        google_start_page = google_start_page + pages_per_loop;
        GoogleSearchResult google_result;
        try{
        google_result = search.doSearch();
        } catch (GoogleSearchFault ex) {
        System.err.println("GoogleSearch.processTextQuery error: the call to the Google Web APIs failed:" + ex.toString());
        // add the error to the result
        return result;
        }
        if (j==0) {
        
        //Total amount of documents Google Search returned
        // only need to do this on the first loop
        
        long numdocs_matched = google_result.getEstimatedTotalResultsCount();       
        GSXML.addMetadata(this.doc, query_metadata_list, "numDocsMatched", ""+numdocs_matched);
        
        }
        GoogleSearchResultElement[] details = google_result.getResultElements();
        for (int i=0; i<details.length; i++){
        Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
        doc_node_list.appendChild(doc_node);
        
        Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
        doc_node.appendChild(metadata_list);
        
        String google_url = details[i].getURL();  
        String google_title = details[i].getTitle();
        String google_snippet = details[i].getSnippet();
        if (google_url !=null) {
            GSXML.addMetadata(this.doc, metadata_list, "URL", google_url);
        }
        if (google_title != null) {
            GSXML.addMetadata(this.doc, metadata_list, "Title", google_title);
        }
        if (google_snippet != null) {
            GSXML.addMetadata(this.doc, metadata_list, "Snippet", google_snippet);
        }
        }
    } // for each loop
    return result;
    }
    
    protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
        index_ids.add("idx");
    index_names.add("Google main index");
    }
    
}