source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/GoogleSearch.java.tmp@ 29558

Last change on this file since 29558 was 29558, checked in by kjdon, 9 years ago

work around does_paging, does_chunking. only add in maxdocs, hitsperpage params if the service actually uses them. lucnee/solr, don't use maxdocs any more. I haven't had a chance to clean up the changes, but I need to commit, so there may be extraneous debug statements still here.

  • Property svn:keywords set to Author Date Id Revision
File size: 7.0 KB
Line 
1package org.greenstone.gsdl3.service;
2
3// Greenstone classes
4import org.greenstone.gsdl3.util.*;
5
6//Google Web Services API classes
7//import com.google.soap.search.GoogleSearch;
8import com.google.soap.search.GoogleSearchFault;
9import com.google.soap.search.GoogleSearchResult;
10import com.google.soap.search.GoogleSearchResultElement;
11import com.google.soap.search.GoogleSearchDirectoryCategory;
12
13// XML classes
14import org.w3c.dom.Element;
15import org.w3c.dom.Document;
16import org.w3c.dom.NodeList;
17
18//Java classes
19import java.util.ArrayList;
20import java.util.HashMap;
21import java.io.File;
22import java.io.InputStream;
23import java.io.BufferedReader;
24import java.io.InputStreamReader;
25import java.io.IOException;
26import java.net.HttpURLConnection;
27import java.net.URLConnection;
28import java.net.URL;
29import java.net.Authenticator;
30import java.net.PasswordAuthentication;
31import java.net.MalformedURLException;
32import java.lang.Object;
33
34import org.apache.log4j.*;
35
36/**
37 *
38 *
39 */
40
41public class GoogleSearch
42 extends AbstractSearch {
43
44 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.GoogleSearch.class.getName());
45
46 //Parameters connect to Proxy Server
47 private boolean using_proxy = false;
48 private String proxy_host = null;
49 private int proxy_port;
50 private String proxy_user = null;
51 private char [] proxy_passwd = null;
52
53 // google key
54 private String client_key = null;
55
56 public GoogleSearch()
57 {
58 does_paging = true;
59 does_chunking = false;
60 QUERY_SERVICE = "TextQuery";
61 }
62
63 //Configure GoogleSearch Service
64 public boolean configure(Element info, Element extra_info)
65 {
66 if (!super.configure(info, extra_info)) {
67 return false;
68 }
69 logger.info("Configuring GoogleSearch");
70 Element server_elem = (Element)GSXML.getChildByTagName(info, "googleServer");
71 if (server_elem == null) {
72 logger.error("no googleServer element found");
73 return false;
74 }
75
76 client_key = server_elem.getAttribute("key");
77 if (client_key.equals("")) {
78 logger.error("no client_key for the googleServer element");
79 return false;
80 }
81
82
83 // are we behind a proxy??
84 // all the details should have been set up by the Message Router
85 proxy_host = System.getProperty("http.proxyHost");
86 if (proxy_host != null && !proxy_host.equals("")) {
87 using_proxy = true;
88 try {
89 proxy_port = Integer.parseInt(System.getProperty("http.proxyPort").trim());
90 } catch (Exception e) {
91 logger.error("couldn't get proxy port, defaulting to 80");
92 proxy_port = 80;
93 }
94 PasswordAuthentication pa = Authenticator.requestPasswordAuthentication(proxy_host, null, proxy_port, "http", "", null);
95 proxy_user = pa.getUserName();
96 proxy_passwd = pa.getPassword();
97 }
98 return true;
99 }
100
101 /** Process a text query - implemented by concrete subclasses */
102 protected Element processTextQuery(Element request) {
103
104 //Connect to Google Web API service
105 com.google.soap.search.GoogleSearch search = new com.google.soap.search.GoogleSearch();
106
107 // Set mandatory attributes
108 search.setKey(client_key);
109
110 // proxy??
111 if (using_proxy) {
112 search.setProxyHost(proxy_host);
113 search.setProxyPort(proxy_port);
114 search.setProxyUserName(proxy_user);
115 search.setProxyPassword(new String(proxy_passwd));
116 }
117
118 //set optional attributes
119 search.setSafeSearch(true);
120
121 // Create a new (empty) result message
122 Document result_doc = XMLConverter.newDOM();
123 Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
124 result.setAttribute(GSXML.FROM_ATT, QUERY_SERVICE);
125 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
126 Element doc_node_list = result_doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
127 result.appendChild(doc_node_list);
128 Element query_metadata_list = result_doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
129 result.appendChild(query_metadata_list);
130
131 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
132 if (param_list == null) {
133 logger.error("TextQuery request had no paramList.");
134 return result; // Return the empty result
135 }
136
137 // Process the request parameters
138 HashMap params = GSXML.extractParams(param_list, false);
139
140 // Make sure a query has been specified
141 String query = (String) params.get(QUERY_PARAM);
142 if (query == null || query.equals("")) {
143 logger.error("TextQuery request had no query string.");
144 return result; // Return the empty result
145 }
146 // tidy whitespace
147 query = query.replaceAll("\\s+", "+");
148
149 search.setQueryString(query);
150
151 //Check hits_per_page
152 int hits_per_page;
153 try {
154 hits_per_page = Integer.parseInt(((String)params.get(HITS_PER_PAGE_PARAM)).trim());
155 } catch (Exception e) {
156 logger.error("couldn't get hits per page param, defaulting to 10");
157 hits_per_page = 10;
158 }
159 //Check the start_page number
160 int start_page;
161 try {
162 start_page = Integer.parseInt(((String) params.get(START_PAGE_PARAM)).trim());
163 } catch (Exception e) {
164 logger.error("couldn't get start page param, defaulting to 1");
165 start_page = 1;
166 }
167
168 //Invoke Actual Search
169
170 // Google only allows 10 hits per request
171 int loop = hits_per_page/10;
172 int remainder = hits_per_page%10;
173 int google_start_page = (start_page-1)*hits_per_page;
174 int pages_per_loop;
175 for (int j=0; j < loop; j++){
176 if (j < (loop-1) || remainder == 0) {
177 pages_per_loop = 10;
178 } else {
179 pages_per_loop = remainder;
180 }
181 search.setMaxResults(pages_per_loop);
182 search.setStartResult(google_start_page);
183 google_start_page = google_start_page + pages_per_loop;
184 GoogleSearchResult google_result;
185 try{
186 google_result = search.doSearch();
187 } catch (GoogleSearchFault ex) {
188 logger.error("the call to the Google Web APIs failed:" + ex.toString());
189 // add the error to the result
190 return result;
191 }
192 if (j==0) {
193
194 //Total amount of documents Google Search returned
195 // only need to do this on the first loop
196
197 long numdocs_matched = google_result.getEstimatedTotalResultsCount();
198 GSXML.addMetadata(query_metadata_list, "numDocsMatched", ""+numdocs_matched);
199
200 }
201 GoogleSearchResultElement[] details = google_result.getResultElements();
202 for (int i=0; i<details.length; i++){
203 Element doc_node = result_doc.createElement(GSXML.DOC_NODE_ELEM);
204 doc_node_list.appendChild(doc_node);
205
206 Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
207 doc_node.appendChild(metadata_list);
208
209 String google_url = details[i].getURL();
210 String google_title = details[i].getTitle();
211 String google_snippet = details[i].getSnippet();
212 if (google_url !=null) {
213 GSXML.addMetadata(metadata_list, "URL", google_url);
214 }
215 if (google_title != null) {
216 GSXML.addMetadata(metadata_list, "Title", google_title);
217 }
218 if (google_snippet != null) {
219 GSXML.addMetadata(metadata_list, "Snippet", google_snippet);
220 }
221 }
222 } // for each loop
223 return result;
224 }
225
226 protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
227 index_ids.add("idx");
228 index_names.add("Google main index");
229 }
230
231}
Note: See TracBrowser for help on using the repository browser.