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 | import org.apache.log4j.*;
|
---|
35 |
|
---|
36 | /**
|
---|
37 | *
|
---|
38 | * @author <a href="mailto:[email protected]">Chi-Yu Huang</a>
|
---|
39 | *
|
---|
40 | */
|
---|
41 |
|
---|
42 | public class GoogleSearch
|
---|
43 | extends AbstractSearch {
|
---|
44 |
|
---|
45 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.GoogleSearch.class.getName());
|
---|
46 |
|
---|
47 | //Parameters connect to Proxy Server
|
---|
48 | private boolean using_proxy = false;
|
---|
49 | private String proxy_host = null;
|
---|
50 | private int proxy_port;
|
---|
51 | private String proxy_user = null;
|
---|
52 | private char [] proxy_passwd = null;
|
---|
53 |
|
---|
54 | // google key
|
---|
55 | private String client_key = null;
|
---|
56 |
|
---|
57 | public GoogleSearch()
|
---|
58 | {
|
---|
59 | }
|
---|
60 |
|
---|
61 | //Configure GoogleSearch Service
|
---|
62 | public boolean configure(Element info, Element extra_info)
|
---|
63 | {
|
---|
64 | if (!super.configure(info, extra_info)) {
|
---|
65 | return false;
|
---|
66 | }
|
---|
67 | logger.info("Configuring GoogleSearch");
|
---|
68 | Element server_elem = (Element)GSXML.getChildByTagName(info, "googleServer");
|
---|
69 | if (server_elem == null) {
|
---|
70 | logger.error("no googleServer element found");
|
---|
71 | return false;
|
---|
72 | }
|
---|
73 |
|
---|
74 | client_key = server_elem.getAttribute("key");
|
---|
75 | if (client_key.equals("")) {
|
---|
76 | logger.error("no client_key for the googleServer element");
|
---|
77 | return false;
|
---|
78 | }
|
---|
79 |
|
---|
80 | does_paging = true;
|
---|
81 | does_chunking = false;
|
---|
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 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
123 | result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
|
---|
124 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
125 | Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
126 | result.appendChild(doc_node_list);
|
---|
127 | Element query_metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
128 | result.appendChild(query_metadata_list);
|
---|
129 |
|
---|
130 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
131 | if (param_list == null) {
|
---|
132 | logger.error("TextQuery request had no paramList.");
|
---|
133 | return result; // Return the empty result
|
---|
134 | }
|
---|
135 |
|
---|
136 | // Process the request parameters
|
---|
137 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
138 |
|
---|
139 | // Make sure a query has been specified
|
---|
140 | String query = (String) params.get(QUERY_PARAM);
|
---|
141 | if (query == null || query.equals("")) {
|
---|
142 | logger.error("TextQuery request had no query string.");
|
---|
143 | return result; // Return the empty result
|
---|
144 | }
|
---|
145 | // tidy whitespace
|
---|
146 | query = query.replaceAll("\\s+", "+");
|
---|
147 |
|
---|
148 | search.setQueryString(query);
|
---|
149 |
|
---|
150 | //Check hits_per_page
|
---|
151 | int hits_per_page;
|
---|
152 | try {
|
---|
153 | hits_per_page = Integer.parseInt(((String)params.get(HITS_PER_PAGE_PARAM)).trim());
|
---|
154 | } catch (Exception e) {
|
---|
155 | logger.error("couldn't get hits per page param, defaulting to 10");
|
---|
156 | hits_per_page = 10;
|
---|
157 | }
|
---|
158 | //Check the start_page number
|
---|
159 | int start_page;
|
---|
160 | try {
|
---|
161 | start_page = Integer.parseInt(((String) params.get(START_PAGE_PARAM)).trim());
|
---|
162 | } catch (Exception e) {
|
---|
163 | logger.error("couldn't get start page param, defaulting to 1");
|
---|
164 | start_page = 1;
|
---|
165 | }
|
---|
166 |
|
---|
167 | //Invoke Actual Search
|
---|
168 |
|
---|
169 | // Google only allows 10 hits per request
|
---|
170 | int loop = hits_per_page/10;
|
---|
171 | int remainder = hits_per_page%10;
|
---|
172 | int google_start_page = (start_page-1)*hits_per_page;
|
---|
173 | int pages_per_loop;
|
---|
174 | for (int j=0; j < loop; j++){
|
---|
175 | if (j < (loop-1) || remainder == 0) {
|
---|
176 | pages_per_loop = 10;
|
---|
177 | } else {
|
---|
178 | pages_per_loop = remainder;
|
---|
179 | }
|
---|
180 | search.setMaxResults(pages_per_loop);
|
---|
181 | search.setStartResult(google_start_page);
|
---|
182 | google_start_page = google_start_page + pages_per_loop;
|
---|
183 | GoogleSearchResult google_result;
|
---|
184 | try{
|
---|
185 | google_result = search.doSearch();
|
---|
186 | } catch (GoogleSearchFault ex) {
|
---|
187 | logger.error("the call to the Google Web APIs failed:" + ex.toString());
|
---|
188 | // add the error to the result
|
---|
189 | return result;
|
---|
190 | }
|
---|
191 | if (j==0) {
|
---|
192 |
|
---|
193 | //Total amount of documents Google Search returned
|
---|
194 | // only need to do this on the first loop
|
---|
195 |
|
---|
196 | long numdocs_matched = google_result.getEstimatedTotalResultsCount();
|
---|
197 | GSXML.addMetadata(this.doc, query_metadata_list, "numDocsMatched", ""+numdocs_matched);
|
---|
198 |
|
---|
199 | }
|
---|
200 | GoogleSearchResultElement[] details = google_result.getResultElements();
|
---|
201 | for (int i=0; i<details.length; i++){
|
---|
202 | Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
203 | doc_node_list.appendChild(doc_node);
|
---|
204 |
|
---|
205 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
206 | doc_node.appendChild(metadata_list);
|
---|
207 |
|
---|
208 | String google_url = details[i].getURL();
|
---|
209 | String google_title = details[i].getTitle();
|
---|
210 | String google_snippet = details[i].getSnippet();
|
---|
211 | if (google_url !=null) {
|
---|
212 | GSXML.addMetadata(this.doc, metadata_list, "URL", google_url);
|
---|
213 | }
|
---|
214 | if (google_title != null) {
|
---|
215 | GSXML.addMetadata(this.doc, metadata_list, "Title", google_title);
|
---|
216 | }
|
---|
217 | if (google_snippet != null) {
|
---|
218 | GSXML.addMetadata(this.doc, metadata_list, "Snippet", google_snippet);
|
---|
219 | }
|
---|
220 | }
|
---|
221 | } // for each loop
|
---|
222 | return result;
|
---|
223 | }
|
---|
224 |
|
---|
225 | protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
|
---|
226 | index_ids.add("idx");
|
---|
227 | index_names.add("Google main index");
|
---|
228 | }
|
---|
229 |
|
---|
230 | }
|
---|