[9304] | 1 | package org.greenstone.gsdl3.service;
|
---|
| 2 |
|
---|
| 3 | // Greenstone classes
|
---|
| 4 | import org.greenstone.gsdl3.util.*;
|
---|
| 5 |
|
---|
| 6 | //Google Web Services API classes
|
---|
| 7 | //import com.google.soap.search.GoogleSearch;
|
---|
| 8 | import com.google.soap.search.GoogleSearchFault;
|
---|
| 9 | import com.google.soap.search.GoogleSearchResult;
|
---|
| 10 | import com.google.soap.search.GoogleSearchResultElement;
|
---|
| 11 | import com.google.soap.search.GoogleSearchDirectoryCategory;
|
---|
| 12 |
|
---|
| 13 | // XML classes
|
---|
| 14 | import org.w3c.dom.Element;
|
---|
| 15 | import org.w3c.dom.Document;
|
---|
| 16 | import org.w3c.dom.NodeList;
|
---|
| 17 |
|
---|
| 18 | //Java classes
|
---|
| 19 | import java.util.ArrayList;
|
---|
| 20 | import java.util.HashMap;
|
---|
| 21 | import java.io.File;
|
---|
| 22 | import java.io.InputStream;
|
---|
| 23 | import java.io.BufferedReader;
|
---|
| 24 | import java.io.InputStreamReader;
|
---|
| 25 | import java.io.IOException;
|
---|
| 26 | import java.net.HttpURLConnection;
|
---|
| 27 | import java.net.URLConnection;
|
---|
| 28 | import java.net.URL;
|
---|
| 29 | import java.net.Authenticator;
|
---|
| 30 | import java.net.PasswordAuthentication;
|
---|
| 31 | import java.net.MalformedURLException;
|
---|
| 32 | import java.lang.Object;
|
---|
| 33 |
|
---|
| 34 |
|
---|
| 35 | /**
|
---|
| 36 | *
|
---|
| 37 | * @author <a href="mailto:[email protected]">Chi-Yu Huang</a>
|
---|
| 38 | *
|
---|
| 39 | */
|
---|
| 40 |
|
---|
| 41 | public class GoogleSearch
|
---|
| 42 | extends AbstractSearch {
|
---|
| 43 |
|
---|
| 44 | //Parameters connect to Proxy Server
|
---|
| 45 | private boolean using_proxy = false;
|
---|
| 46 | private String proxy_host = null;
|
---|
| 47 | private int proxy_port;
|
---|
| 48 | private String proxy_user = null;
|
---|
| 49 | private char [] proxy_passwd = null;
|
---|
| 50 |
|
---|
| 51 | // google key
|
---|
| 52 | private String client_key = null;
|
---|
| 53 |
|
---|
| 54 | public GoogleSearch()
|
---|
| 55 | {
|
---|
| 56 | }
|
---|
| 57 |
|
---|
| 58 | //Configure GoogleSearch Service
|
---|
| 59 | public boolean configure(Element info, Element extra_info)
|
---|
| 60 | {
|
---|
| 61 | if (!super.configure(info, extra_info)) {
|
---|
| 62 | return false;
|
---|
| 63 | }
|
---|
[11228] | 64 | System.err.println("Configuring GoogleSearch");
|
---|
[9304] | 65 | Element server_elem = (Element)GSXML.getChildByTagName(info, "googleServer");
|
---|
| 66 | if (server_elem == null) {
|
---|
| 67 | System.err.println("GoogleSearch.configure error: no googleServer element found");
|
---|
| 68 | return false;
|
---|
| 69 | }
|
---|
| 70 |
|
---|
| 71 | client_key = server_elem.getAttribute("key");
|
---|
| 72 | if (client_key.equals("")) {
|
---|
| 73 | System.err.println("GoogleSearch.configure error: no client_key for the googleServer element");
|
---|
| 74 | return false;
|
---|
| 75 | }
|
---|
| 76 |
|
---|
| 77 | does_paging = true;
|
---|
| 78 | does_chunking = false;
|
---|
| 79 |
|
---|
| 80 | // are we behind a proxy??
|
---|
| 81 | // all the details should have been set up by the Message Router
|
---|
| 82 | proxy_host = System.getProperty("http.proxyHost");
|
---|
| 83 | if (proxy_host != null && !proxy_host.equals("")) {
|
---|
| 84 | using_proxy = true;
|
---|
| 85 | try {
|
---|
| 86 | proxy_port = Integer.parseInt(System.getProperty("http.proxyPort").trim());
|
---|
| 87 | } catch (Exception e) {
|
---|
| 88 | System.err.println("GoogleSearch.configure error: couldn't get proxy port, defaulting to 80");
|
---|
| 89 | proxy_port = 80;
|
---|
| 90 | }
|
---|
| 91 | PasswordAuthentication pa = Authenticator.requestPasswordAuthentication(proxy_host, null, proxy_port, "http", "", null);
|
---|
| 92 | proxy_user = pa.getUserName();
|
---|
| 93 | proxy_passwd = pa.getPassword();
|
---|
| 94 | }
|
---|
| 95 | return true;
|
---|
| 96 | }
|
---|
| 97 |
|
---|
| 98 | /** Process a text query - implemented by concrete subclasses */
|
---|
| 99 | protected Element processTextQuery(Element request) {
|
---|
| 100 |
|
---|
| 101 | //Connect to Google Web API service
|
---|
| 102 | com.google.soap.search.GoogleSearch search = new com.google.soap.search.GoogleSearch();
|
---|
| 103 |
|
---|
| 104 | // Set mandatory attributes
|
---|
| 105 | search.setKey(client_key);
|
---|
| 106 |
|
---|
| 107 | // proxy??
|
---|
| 108 | if (using_proxy) {
|
---|
| 109 | search.setProxyHost(proxy_host);
|
---|
| 110 | search.setProxyPort(proxy_port);
|
---|
| 111 | search.setProxyUserName(proxy_user);
|
---|
| 112 | search.setProxyPassword(new String(proxy_passwd));
|
---|
| 113 | }
|
---|
| 114 |
|
---|
| 115 | //set optional attributes
|
---|
| 116 | search.setSafeSearch(true);
|
---|
| 117 |
|
---|
| 118 | // Create a new (empty) result message
|
---|
| 119 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
| 120 | result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
|
---|
| 121 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
| 122 | Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 123 | result.appendChild(doc_node_list);
|
---|
| 124 | Element query_metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 125 | result.appendChild(query_metadata_list);
|
---|
| 126 |
|
---|
| 127 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 128 | if (param_list == null) {
|
---|
| 129 | System.err.println("GoogleSearch.processTextQuery Error:: TextQuery request had no paramList.");
|
---|
| 130 | return result; // Return the empty result
|
---|
| 131 | }
|
---|
| 132 |
|
---|
| 133 | // Process the request parameters
|
---|
| 134 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
| 135 |
|
---|
| 136 | // Make sure a query has been specified
|
---|
| 137 | String query = (String) params.get(QUERY_PARAM);
|
---|
| 138 | if (query == null || query.equals("")) {
|
---|
| 139 | System.err.println("GoogleSearch.processTextQuery Error: TextQuery request had no query string.");
|
---|
| 140 | return result; // Return the empty result
|
---|
| 141 | }
|
---|
| 142 | // tidy whitespace
|
---|
| 143 | query = query.replaceAll("\\s+", "+");
|
---|
| 144 |
|
---|
| 145 | search.setQueryString(query);
|
---|
| 146 |
|
---|
| 147 | //Check hits_per_page
|
---|
| 148 | int hits_per_page;
|
---|
| 149 | try {
|
---|
| 150 | hits_per_page = Integer.parseInt(((String)params.get(HITS_PER_PAGE_PARAM)).trim());
|
---|
| 151 | } catch (Exception e) {
|
---|
| 152 | System.err.println("GoogleSearch.processTextQuery error: couldn't get hits per page param, defaulting to 10");
|
---|
| 153 | hits_per_page = 10;
|
---|
| 154 | }
|
---|
| 155 | //Check the start_page number
|
---|
| 156 | int start_page;
|
---|
| 157 | try {
|
---|
| 158 | start_page = Integer.parseInt(((String) params.get(START_PAGE_PARAM)).trim());
|
---|
| 159 | } catch (Exception e) {
|
---|
| 160 | System.err.println("GoogleSearch.processTextQuery error: couldn't get start page param, defaulting to 1");
|
---|
| 161 | start_page = 1;
|
---|
| 162 | }
|
---|
| 163 |
|
---|
| 164 | //Invoke Actual Search
|
---|
| 165 |
|
---|
| 166 | // Google only allows 10 hits per request
|
---|
| 167 | int loop = hits_per_page/10;
|
---|
| 168 | int remainder = hits_per_page%10;
|
---|
| 169 | int google_start_page = (start_page-1)*hits_per_page;
|
---|
| 170 | int pages_per_loop;
|
---|
| 171 | for (int j=0; j < loop; j++){
|
---|
| 172 | if (j < (loop-1) || remainder == 0) {
|
---|
| 173 | pages_per_loop = 10;
|
---|
| 174 | } else {
|
---|
| 175 | pages_per_loop = remainder;
|
---|
| 176 | }
|
---|
| 177 | search.setMaxResults(pages_per_loop);
|
---|
| 178 | search.setStartResult(google_start_page);
|
---|
| 179 | google_start_page = google_start_page + pages_per_loop;
|
---|
| 180 | GoogleSearchResult google_result;
|
---|
| 181 | try{
|
---|
| 182 | google_result = search.doSearch();
|
---|
| 183 | } catch (GoogleSearchFault ex) {
|
---|
| 184 | System.err.println("GoogleSearch.processTextQuery error: the call to the Google Web APIs failed:" + ex.toString());
|
---|
| 185 | // add the error to the result
|
---|
| 186 | return result;
|
---|
| 187 | }
|
---|
| 188 | if (j==0) {
|
---|
| 189 |
|
---|
| 190 | //Total amount of documents Google Search returned
|
---|
| 191 | // only need to do this on the first loop
|
---|
| 192 |
|
---|
| 193 | long numdocs_matched = google_result.getEstimatedTotalResultsCount();
|
---|
| 194 | GSXML.addMetadata(this.doc, query_metadata_list, "numDocsMatched", ""+numdocs_matched);
|
---|
| 195 |
|
---|
| 196 | }
|
---|
| 197 | GoogleSearchResultElement[] details = google_result.getResultElements();
|
---|
| 198 | for (int i=0; i<details.length; i++){
|
---|
| 199 | Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
| 200 | doc_node_list.appendChild(doc_node);
|
---|
| 201 |
|
---|
| 202 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 203 | doc_node.appendChild(metadata_list);
|
---|
| 204 |
|
---|
| 205 | String google_url = details[i].getURL();
|
---|
| 206 | String google_title = details[i].getTitle();
|
---|
| 207 | String google_snippet = details[i].getSnippet();
|
---|
| 208 | if (google_url !=null) {
|
---|
| 209 | GSXML.addMetadata(this.doc, metadata_list, "URL", google_url);
|
---|
| 210 | }
|
---|
| 211 | if (google_title != null) {
|
---|
| 212 | GSXML.addMetadata(this.doc, metadata_list, "Title", google_title);
|
---|
| 213 | }
|
---|
| 214 | if (google_snippet != null) {
|
---|
| 215 | GSXML.addMetadata(this.doc, metadata_list, "Snippet", google_snippet);
|
---|
| 216 | }
|
---|
| 217 | }
|
---|
| 218 | } // for each loop
|
---|
| 219 | return result;
|
---|
| 220 | }
|
---|
| 221 |
|
---|
| 222 | protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
|
---|
| 223 | index_ids.add("idx");
|
---|
| 224 | index_names.add("Google main index");
|
---|
| 225 | }
|
---|
| 226 |
|
---|
| 227 | }
|
---|