source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 11014

Last change on this file since 11014 was 11014, checked in by kjdon, 18 years ago

cross collection searching service

  • Property svn:keywords set to Author Date Id Revision
File size: 15.7 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.NodeList;
26
27import java.util.ArrayList;
28import java.util.Map;
29import java.util.Iterator;
30import java.util.Set;
31import java.util.HashMap;
32
33
34/**
35 *
36 * @author <a href="mailto:[email protected]">Katherine Don</a>
37 */
38
39public class CrossCollectionSearch
40 extends ServiceRack {
41
42 protected static final String QUERY_PARAM = "query";
43 protected static final String COLLECTION_PARAM = "collection";
44
45 // the services on offer - these proxy the actual collection ones
46 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
47 protected static final String ADV_QUERY_SERVICE = "AdvTextQuery";
48 protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
49
50 protected String [] coll_ids_list = null;
51 protected String [] coll_names_list = null;
52
53
54 /** constructor */
55 public CrossCollectionSearch()
56 {
57 }
58
59 public boolean configure(Element info, Element extra_info)
60 {
61 // any parameters? colls to include??
62 System.out.println("Configuring CrossCollectionSearch...");
63 // query service
64 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
65 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
66 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
67 this.short_service_info.appendChild(ccs_service);
68
69 // metadata service
70 Element dmr_service = this.doc.createElement(GSXML.SERVICE_ELEM);
71 dmr_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
72 dmr_service.setAttribute(GSXML.NAME_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
73 this.short_service_info.appendChild(dmr_service);
74
75 // add in the default format statement
76 String format_info = "<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;&amp;p.s="+TEXT_QUERY_SERVICE+"&amp;p.c=";
77 if (this.cluster_name!=null) {
78 format_info += this.cluster_name;
79 }
80 format_info += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>, rank:<xsl:value-of select='@rank'/>) </td></gsf:template></format>";
81 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_info).getDocumentElement(), true));
82 return true;
83 }
84
85 protected Element getServiceDescription(String service, String lang, String subset)
86 {
87 if (service.equals(TEXT_QUERY_SERVICE)) {
88
89 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
90 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
91 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
92
93 // display info
94 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
95 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE+".name", lang)));
96 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE+".submit", lang)));
97 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE+".description", lang)));
98 }
99 // param info
100 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
101 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
102 // collection list
103 if (coll_ids_list ==null ) {
104 initCollectionList();
105 }
106 Element param = GSXML.createParameterDescription(this.doc, COLLECTION_PARAM, getTextString("param."+COLLECTION_PARAM, lang), GSXML.PARAM_TYPE_ENUM_MULTI, null, coll_ids_list, coll_names_list);
107 param_list.appendChild(param);
108 // query param
109 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
110 param_list.appendChild(param);
111 ccs_service.appendChild(param_list);
112 }
113
114 ///ystem.err.println("service description="+this.converter.getPrettyString(ccs_service));
115 return ccs_service;
116 }
117 // these ones are probably never called, but put them here just in case
118 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
119 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
120 service_elem.setAttribute(GSXML.NAME_ATT, service);
121 return service_elem;
122
123
124 }
125
126
127 protected Element processTextQuery(Element request) {
128 // Create a new (empty) result message
129 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
130 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
131 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
132
133 String lang = request.getAttribute(GSXML.LANG_ATT);
134 // Get the parameters of the request
135 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
136 if (param_list == null) {
137 System.err.println("Error: TextQuery request had no paramList.");
138 return result; // Return the empty result
139 }
140
141 // get the collection list
142 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
143 String coll_list = GSXML.getValue(coll_param);
144
145 String [] colls = coll_list.split(",");
146 Element query_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
147 // we are sending the same request to each collection - build up the to
148 // attribute for the request
149 StringBuffer to_att = new StringBuffer();
150 for (int i=0; i<colls.length; i++) {
151 if (i>0) {
152 to_att.append(",");
153 }
154 to_att.append(GSPath.appendLink(colls[i], "TextQuery"));
155
156 }
157 // send the query to all colls
158 Element query_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), lang, "");
159 query_message.appendChild(query_request);
160 // should we add params individually?
161 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
162 query_request.appendChild(new_param_list);
163 new_param_list.appendChild(this.doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
164 Element query_result = this.router.process(query_message);
165
166 // gather up the data from each response
167 int numDocsMatched = 0;
168 int numDocsReturned = 0;
169
170 //term info??
171
172 NodeList metadata = query_result.getElementsByTagName(GSXML.METADATA_ELEM);
173 for (int j=0; j<metadata.getLength(); j++) {
174 Element meta = (Element)metadata.item(j);
175 if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsReturned")) {
176 numDocsReturned += Integer.parseInt(GSXML.getValue(meta));
177 } else if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsMatched")) {
178 numDocsMatched += Integer.parseInt(GSXML.getValue(meta));
179 }
180 }
181
182 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
183 result.appendChild(metadata_list);
184 GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", ""+numDocsReturned);
185 //GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numDocsMatched);
186
187 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
188 result.appendChild(doc_node_list);
189
190 NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
191
192 for (int k=0; k<responses.getLength(); k++) {
193 String coll_name = GSPath.getFirstLink(((Element)responses.item(k)).getAttribute(GSXML.FROM_ATT));
194 NodeList nodes = ((Element)responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
195 if (nodes==null || nodes.getLength()==0) continue;
196 Element last_node = null;
197 Element this_node = null;
198 for (int n=0; n<nodes.getLength(); n++) {
199 this_node = (Element)nodes.item(n);
200 this_node.setAttribute("collection", coll_name);
201 if (k==0) {
202
203 doc_node_list.appendChild(this.doc.importNode(this_node, true));
204 } else {
205 if (last_node==null) {
206 last_node = (Element)GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
207 }
208 last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
209 }
210
211 }
212 }
213 return result;
214 }
215
216// protected Element processAdvTextQuery(Element request)
217// {
218
219// }
220 protected boolean initCollectionList() {
221 String lang="en";
222 String uid = "";
223 // first, get the message router info
224 Element coll_list_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
225 Element coll_list_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, "", lang, ""); // uid
226 coll_list_message.appendChild(coll_list_request);
227 ///ystem.err.println("coll list request = "+this.converter.getPrettyString(coll_list_request));
228 Element coll_list_response = (Element)this.router.process(coll_list_message);
229 if (coll_list_response==null) {
230 System.err.println("PageAction Error: couldn't query the message router!");
231 return false;
232 }
233 ///ystem.err.println("coll list response = "+this.converter.getPrettyString(coll_list_response));
234 // second, get some info from each collection. we want the coll name
235 // and whether its got a text query service
236
237 NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
238 // we will send all the requests in a single message
239 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
240 for (int i=0; i<colls.getLength(); i++) {
241 Element c = (Element)colls.item(i);
242 String name = c.getAttribute(GSXML.NAME_ATT);
243 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, lang, uid);
244 metadata_message.appendChild(metadata_request);
245 }
246 ///ystem.err.println("metadata request = "+this.converter.getPrettyString(metadata_message));
247 Element metadata_response = (Element)this.router.process(metadata_message);
248 ///ystem.err.println("metadata response = "+this.converter.getPrettyString(metadata_response));
249
250 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
251 ArrayList valid_colls = new ArrayList();
252 ArrayList valid_coll_names = new ArrayList();
253 for (int i=0; i<coll_responses.getLength(); i++) {
254 Element coll = (Element)coll_responses.item(i);
255 Element service_list = (Element)GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM+GSXML.LIST_MODIFIER);
256 if (service_list==null) continue;
257 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractSearch.TEXT_QUERY_SERVICE
258 if (query_service == null) continue;
259 String coll_id = coll.getAttribute(GSXML.NAME_ATT);
260 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
261 valid_colls.add(coll_id);
262 valid_coll_names.add(coll_name);
263 }
264
265 if (valid_colls.size()==0) {
266 return false;
267 }
268 this.coll_ids_list = new String[1];
269 this.coll_names_list = new String[1];
270 this.coll_ids_list = (String []) valid_colls.toArray(coll_ids_list);
271 this.coll_names_list = (String []) valid_coll_names.toArray(coll_names_list);
272 return true;
273 }
274
275 protected Element processDocumentMetadataRetrieve(Element request) {
276 // Create a new (empty) result message
277 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
278 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
279 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
280
281 String lang = request.getAttribute(GSXML.LANG_ATT);
282 // Get the parameters of the request
283 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
284 if (param_list == null) {
285 System.err.println("Error: DocumentMetadataRetrieve request had no paramList.");
286 return result; // Return the empty result
287 }
288
289 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
290 if (query_doc_list.getLength()==0) {
291 System.err.println("Error: DocumentMetadataRetrieve request had no documentNodes.");
292 return result; // Return the empty result
293 }
294
295 // the resulting doc node list
296 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
297 result.appendChild(result_node_list);
298
299 // get all the metadata params
300 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
301 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
302 new_param_list.appendChild(param);
303
304 // organise the nodes into collection lists
305 HashMap coll_map = new HashMap();
306
307 for (int i=0; i<query_doc_list.getLength(); i++) {
308 Element doc_node = (Element)query_doc_list.item(i);
309 String coll_name = doc_node.getAttribute("collection");
310 Element coll_items = (Element)coll_map.get(coll_name);
311 if (coll_items==null) {
312 coll_items=this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
313 coll_map.put(coll_name, coll_items);
314 }
315 coll_items.appendChild(this.doc.importNode(doc_node, true));
316 }
317
318 // create teh individual requests
319 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
320 Set mapping_set = coll_map.entrySet();
321 Iterator iter = mapping_set.iterator();
322
323 while (iter.hasNext()) {
324 Map.Entry e = (Map.Entry)iter.next();
325 String cname = (String)e.getKey();
326 Element doc_nodes = (Element)e.getValue();
327 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), lang, "");
328 meta_request.appendChild(doc_nodes);
329 meta_request.appendChild(new_param_list.cloneNode(true));
330 meta_request_message.appendChild(meta_request);
331
332 }
333 Element meta_result = this.router.process(meta_request_message);
334
335 // now need to put the doc nodes back in the right order
336 // go through the original list again. keep an element pointer to
337 // the next element in each collections list
338 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
339 for (int i=0; i<meta_responses.getLength(); i++) {
340 String collname = GSPath.getFirstLink(((Element)meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
341 Element first_elem = (Element)GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
342 coll_map.put(collname, first_elem);
343 }
344
345 for (int i=0; i<query_doc_list.getLength(); i++) {
346 Element doc_node = (Element)query_doc_list.item(i);
347 Element new_node = (Element)this.doc.importNode(doc_node, false);
348 result_node_list.appendChild(new_node);
349 String coll_name = doc_node.getAttribute("collection");
350
351 Element meta_elem = (Element)coll_map.get(coll_name);
352 GSXML.mergeMetadataLists(new_node, meta_elem);
353 coll_map.put(coll_name, meta_elem.getNextSibling());
354 }
355 return result;
356 }
357}
358
Note: See TracBrowser for help on using the repository browser.