source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 24394

Last change on this file since 24394 was 24394, checked in by davidb, 13 years ago

Through the audioDB extension we now support a form of content-based audio/music searching. These commited changes reflect this generalization in our Service inheritance hierarchy for searching. Basically, what used to be thought of as a search service implied a *text* search service.
  • Property svn:keywords set to Author Date Id Revision
File size: 16.7 KB
Line 
1/*
2 * CrossCollectionSearch.java
3 * Copyright (C) 2002 New Zealand Digital Library, http://www.nzdl.org
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19package org.greenstone.gsdl3.service;
20
21import org.greenstone.gsdl3.util.GSXML;
22import org.greenstone.gsdl3.util.GSPath;
23
24import org.w3c.dom.Element;
25import org.w3c.dom.Node;
26import org.w3c.dom.NodeList;
27
28import java.util.ArrayList;
29import java.util.Map;
30import java.util.Iterator;
31import java.util.Set;
32import java.util.HashMap;
33
34import org.apache.log4j.*;
35
36/**
37 *
38 * @author <a href="mailto:[email protected]">Katherine Don</a>
39 */
40
41public class CrossCollectionSearch
42 extends ServiceRack {
43
44 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.CrossCollectionSearch.class.getName());
45 protected static final String QUERY_PARAM = "query";
46 protected static final String COLLECTION_PARAM = "collection";
47
48 // the services on offer - these proxy the actual collection ones
49 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
50 protected static final String ADV_QUERY_SERVICE = "AdvTextQuery";
51 protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
52
53 protected String [] coll_ids_list = null;
54 protected String [] coll_ids_list_no_all = null;
55 protected String [] coll_names_list = null;
56
57
58 /** constructor */
59 public CrossCollectionSearch()
60 {
61 }
62
63 public boolean configure(Element info, Element extra_info)
64 {
65 // any parameters? colls to include??
66 logger.info("Configuring CrossCollectionSearch...");
67 // query service
68 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
69 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
70 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
71 this.short_service_info.appendChild(ccs_service);
72
73 // metadata service
74 Element dmr_service = this.doc.createElement(GSXML.SERVICE_ELEM);
75 dmr_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
76 dmr_service.setAttribute(GSXML.NAME_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
77 this.short_service_info.appendChild(dmr_service);
78
79 // get any format info
80 Element format_info = (Element)GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
81 if (format_info != null) {
82 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format_info, true));
83 } else {
84 // add in a default format statement
85 String format_string = "<format xmlns:gsf='http://www.greenstone.org/greenstone3/schema/ConfigFormat' xmlns:xsl='http://www.w3.org/1999/XSL/Transform'><gsf:template match='documentNode'><td><a><xsl:attribute name='href'>?a=d&amp;c=<xsl:value-of select='@collection'/>&amp;d=<xsl:value-of select='@nodeID'/><xsl:if test=\"@nodeType='leaf'\">&amp;sib=1</xsl:if>&amp;dt=<xsl:value-of select='@docType'/>&amp;p.a=q&amp;p.s="+TEXT_QUERY_SERVICE+"&amp;p.c=";
86 if (this.cluster_name!=null) {
87 format_string += this.cluster_name;
88 }
89 format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
90 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
91 }
92 return true;
93 }
94
95 protected Element getServiceDescription(String service, String lang, String subset)
96 {
97 if (service.equals(TEXT_QUERY_SERVICE)) {
98
99 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
100 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
101 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
102
103 // display info
104 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
105 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE+".name", lang)));
106 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE+".submit", lang)));
107 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE+".description", lang)));
108 }
109 // param info
110 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
111 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
112 // collection list
113 if (coll_ids_list ==null ) {
114 initCollectionList();
115 }
116 Element param = GSXML.createParameterDescription(this.doc, COLLECTION_PARAM, getTextString("param."+COLLECTION_PARAM, lang), GSXML.PARAM_TYPE_ENUM_MULTI, "all", coll_ids_list, coll_names_list);
117 param_list.appendChild(param);
118 // query param
119 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
120 param_list.appendChild(param);
121 ccs_service.appendChild(param_list);
122 }
123
124 logger.debug("service description="+this.converter.getPrettyString(ccs_service));
125 return ccs_service;
126 }
127 // these ones are probably never called, but put them here just in case
128 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
129 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
130 service_elem.setAttribute(GSXML.NAME_ATT, service);
131 return service_elem;
132
133
134 }
135
136
137 protected Element processTextQuery(Element request) {
138 // Create a new (empty) result message
139 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
140 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
141 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
142
143 String lang = request.getAttribute(GSXML.LANG_ATT);
144 // Get the parameters of the request
145 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
146 if (param_list == null) {
147 logger.error("TextQuery request had no paramList.");
148 return result; // Return the empty result
149 }
150
151 // get the collection list
152 String [] colls_list = coll_ids_list_no_all;
153 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
154 if (coll_param != null) {
155 String coll_list = GSXML.getValue(coll_param);
156 if (!coll_list.equals("all") && !coll_list.equals("")) {
157 colls_list = coll_list.split(",");
158 }
159 }
160
161 Element query_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
162 // we are sending the same request to each collection - build up the to
163 // attribute for the request
164 StringBuffer to_att = new StringBuffer();
165 for (int i=0; i<colls_list.length; i++) {
166 if (i>0) {
167 to_att.append(",");
168 }
169 to_att.append(GSPath.appendLink(colls_list[i], "TextQuery"));
170
171 }
172 // send the query to all colls
173 Element query_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), lang, "");
174 query_message.appendChild(query_request);
175 // should we add params individually?
176 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
177 query_request.appendChild(new_param_list);
178 new_param_list.appendChild(this.doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
179 Element query_result = (Element)this.router.process(query_message);
180
181 // gather up the data from each response
182 int numDocsMatched = 0;
183 int numDocsReturned = 0;
184
185 //term info??
186
187 NodeList metadata = query_result.getElementsByTagName(GSXML.METADATA_ELEM);
188 for (int j=0; j<metadata.getLength(); j++) {
189 Element meta = (Element)metadata.item(j);
190 if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsReturned")) {
191 numDocsReturned += Integer.parseInt(GSXML.getValue(meta));
192 } else if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsMatched")) {
193 numDocsMatched += Integer.parseInt(GSXML.getValue(meta));
194 }
195 }
196
197 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
198 result.appendChild(metadata_list);
199 GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", ""+numDocsReturned);
200 //GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numDocsMatched);
201
202 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
203 result.appendChild(doc_node_list);
204
205 NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
206
207 for (int k=0; k<responses.getLength(); k++) {
208 String coll_name = GSPath.removeLastLink(((Element)responses.item(k)).getAttribute(GSXML.FROM_ATT));
209 NodeList nodes = ((Element)responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
210 if (nodes==null || nodes.getLength()==0) continue;
211 Element last_node = null;
212 Element this_node = null;
213 for (int n=0; n<nodes.getLength(); n++) {
214 this_node = (Element)nodes.item(n);
215 this_node.setAttribute("collection", coll_name);
216 if (k==0) {
217
218 doc_node_list.appendChild(this.doc.importNode(this_node, true));
219 } else {
220 if (last_node==null) {
221 last_node = (Element)GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
222 }
223 last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
224 }
225
226 }
227 }
228 return result;
229 }
230
231// protected Element processAdvTextQuery(Element request)
232// {
233
234// }
235 protected boolean initCollectionList() {
236 String lang="en";
237 String uid = "";
238 // first, get the message router info
239 Element coll_list_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
240 Element coll_list_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, "", lang, ""); // uid
241 coll_list_message.appendChild(coll_list_request);
242 logger.debug("coll list request = "+this.converter.getPrettyString(coll_list_request));
243 Element coll_list_response = (Element)this.router.process(coll_list_message);
244 if (coll_list_response==null) {
245 logger.error("couldn't query the message router!");
246 return false;
247 }
248 logger.debug("coll list response = "+this.converter.getPrettyString(coll_list_response));
249 // second, get some info from each collection. we want the coll name
250 // and whether its got a text query service
251
252 NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
253 // we will send all the requests in a single message
254 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
255 for (int i=0; i<colls.getLength(); i++) {
256 Element c = (Element)colls.item(i);
257 String name = c.getAttribute(GSXML.NAME_ATT);
258 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, lang, uid);
259 metadata_message.appendChild(metadata_request);
260 }
261 logger.debug("metadata request = "+this.converter.getPrettyString(metadata_message));
262 Element metadata_response = (Element)this.router.process(metadata_message);
263 logger.debug("metadata response = "+this.converter.getPrettyString(metadata_response));
264 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
265 ArrayList valid_colls = new ArrayList();
266 ArrayList valid_coll_names = new ArrayList();
267 for (int i=0; i<coll_responses.getLength(); i++) {
268 Element response = (Element)coll_responses.item(i);
269 Element coll = (Element)GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
270 Element service_list = (Element)GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM+GSXML.LIST_MODIFIER);
271 if (service_list==null) continue;
272 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
273 if (query_service == null) continue;
274 // use the name of the response in case we are talking to a remote collection, not the name of the collection.
275 String coll_id = response.getAttribute(GSXML.FROM_ATT);
276 String coll_name = coll_id+": "+GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, "en", "en"); // just use english for now until we do some caching or something
277 valid_colls.add(coll_id);
278 valid_coll_names.add(coll_name);
279 }
280
281 if (valid_colls.size()==0) {
282 return false;
283 }
284 // ids no all has the list without 'all' option.
285 this.coll_ids_list_no_all = new String[1];
286 this.coll_ids_list_no_all = (String []) valid_colls.toArray(coll_ids_list_no_all);
287
288 valid_colls.add(0, "all");
289 valid_coll_names.add(0, getTextString("param."+COLLECTION_PARAM+".all", "en" ));
290 this.coll_ids_list = new String[1];
291 this.coll_names_list = new String[1];
292 this.coll_ids_list = (String []) valid_colls.toArray(coll_ids_list);
293 this.coll_names_list = (String []) valid_coll_names.toArray(coll_names_list);
294 return true;
295 }
296
297 protected Element processDocumentMetadataRetrieve(Element request) {
298 // Create a new (empty) result message
299 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
300 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
301 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
302
303 String lang = request.getAttribute(GSXML.LANG_ATT);
304 // Get the parameters of the request
305 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
306 if (param_list == null) {
307 logger.error("DocumentMetadataRetrieve request had no paramList.");
308 return result; // Return the empty result
309 }
310
311 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
312 if (query_doc_list.getLength()==0) {
313 logger.error("DocumentMetadataRetrieve request had no documentNodes.");
314 return result; // Return the empty result
315 }
316
317 // the resulting doc node list
318 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
319 result.appendChild(result_node_list);
320
321 // get all the metadata params
322 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
323 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
324 new_param_list.appendChild(param);
325
326 // organise the nodes into collection lists
327 HashMap coll_map = new HashMap();
328
329 for (int i=0; i<query_doc_list.getLength(); i++) {
330 Element doc_node = (Element)query_doc_list.item(i);
331 String coll_name = doc_node.getAttribute("collection");
332 Element coll_items = (Element)coll_map.get(coll_name);
333 if (coll_items==null) {
334 coll_items=this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
335 coll_map.put(coll_name, coll_items);
336 }
337 coll_items.appendChild(this.doc.importNode(doc_node, true));
338 }
339
340 // create teh individual requests
341 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
342 Set mapping_set = coll_map.entrySet();
343 Iterator iter = mapping_set.iterator();
344
345 while (iter.hasNext()) {
346 Map.Entry e = (Map.Entry)iter.next();
347 String cname = (String)e.getKey();
348 Element doc_nodes = (Element)e.getValue();
349 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), lang, "");
350 meta_request.appendChild(doc_nodes);
351 meta_request.appendChild(new_param_list.cloneNode(true));
352 meta_request_message.appendChild(meta_request);
353
354 }
355
356 Node meta_result_node = this.router.process(meta_request_message);
357 Element meta_result = this.converter.nodeToElement(meta_result_node);
358
359 // now need to put the doc nodes back in the right order
360 // go through the original list again. keep an element pointer to
361 // the next element in each collections list
362 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
363 for (int i=0; i<meta_responses.getLength(); i++) {
364 String collname = GSPath.removeLastLink(((Element)meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
365 Element first_elem = (Element)GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
366 coll_map.put(collname, first_elem);
367 }
368
369 for (int i=0; i<query_doc_list.getLength(); i++) {
370 Element doc_node = (Element)query_doc_list.item(i);
371 Element new_node = (Element)this.doc.importNode(doc_node, false);
372 result_node_list.appendChild(new_node);
373 String coll_name = doc_node.getAttribute("collection");
374
375 Element meta_elem = (Element)coll_map.get(coll_name);
376 GSXML.mergeMetadataLists(new_node, meta_elem);
377 coll_map.put(coll_name, meta_elem.getNextSibling());
378 }
379 return result;
380 }
381}
382
Note: See TracBrowser for help on using the repository browser.