source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 11229

Last change on this file since 11229 was 11229, checked in by kjdon, 18 years ago

first look for a format from the serviceRack element before using the default. Also added an 'all' opiton to the collection list

  • Property svn:keywords set to Author Date Id Revision
File size: 16.2 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 // get any format info
76 Element format_info = (Element)GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
77 if (format_info != null) {
78 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format_info, true));
79 } else {
80 // add in a default format statement
81 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;&amp;p.s="+TEXT_QUERY_SERVICE+"&amp;p.c=";
82 if (this.cluster_name!=null) {
83 format_string += this.cluster_name;
84 }
85 format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
86 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
87 }
88 return true;
89 }
90
91 protected Element getServiceDescription(String service, String lang, String subset)
92 {
93 if (service.equals(TEXT_QUERY_SERVICE)) {
94
95 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
96 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
97 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
98
99 // display info
100 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
101 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE+".name", lang)));
102 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE+".submit", lang)));
103 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE+".description", lang)));
104 }
105 // param info
106 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
107 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
108 // collection list
109 if (coll_ids_list ==null ) {
110 initCollectionList();
111 }
112 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);
113 param_list.appendChild(param);
114 // query param
115 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
116 param_list.appendChild(param);
117 ccs_service.appendChild(param_list);
118 }
119
120 ///ystem.err.println("service description="+this.converter.getPrettyString(ccs_service));
121 return ccs_service;
122 }
123 // these ones are probably never called, but put them here just in case
124 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
125 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
126 service_elem.setAttribute(GSXML.NAME_ATT, service);
127 return service_elem;
128
129
130 }
131
132
133 protected Element processTextQuery(Element request) {
134 // Create a new (empty) result message
135 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
136 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
137 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
138
139 String lang = request.getAttribute(GSXML.LANG_ATT);
140 // Get the parameters of the request
141 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
142 if (param_list == null) {
143 System.err.println("Error: TextQuery request had no paramList.");
144 return result; // Return the empty result
145 }
146
147 // get the collection list
148 String [] colls_list;
149 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
150 if (coll_param == null) {
151 colls_list = coll_ids_list;
152 } else {
153 String coll_list = GSXML.getValue(coll_param);
154 if (coll_list.equals("all") || coll_list.equals("")) {
155 colls_list = coll_ids_list;
156 } else {
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 = 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.getFirstLink(((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 ///ystem.err.println("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 System.err.println("PageAction Error: couldn't query the message router!");
246 return false;
247 }
248 ///ystem.err.println("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 ///ystem.err.println("metadata request = "+this.converter.getPrettyString(metadata_message));
262 Element metadata_response = (Element)this.router.process(metadata_message);
263 ///ystem.err.println("metadata response = "+this.converter.getPrettyString(metadata_response));
264
265 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
266 ArrayList valid_colls = new ArrayList();
267 ArrayList valid_coll_names = new ArrayList();
268 for (int i=0; i<coll_responses.getLength(); i++) {
269 Element coll = (Element)coll_responses.item(i);
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 AbstractSearch.TEXT_QUERY_SERVICE
273 if (query_service == null) continue;
274 String coll_id = coll.getAttribute(GSXML.NAME_ATT);
275 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
276 valid_colls.add(coll_id);
277 valid_coll_names.add(coll_name);
278 }
279
280 if (valid_colls.size()==0) {
281 return false;
282 }
283 valid_colls.add(0, "all");
284 valid_coll_names.add(0, getTextString("param."+COLLECTION_PARAM+".all", "en" ));
285 this.coll_ids_list = new String[1];
286 this.coll_names_list = new String[1];
287 this.coll_ids_list = (String []) valid_colls.toArray(coll_ids_list);
288 this.coll_names_list = (String []) valid_coll_names.toArray(coll_names_list);
289 return true;
290 }
291
292 protected Element processDocumentMetadataRetrieve(Element request) {
293 // Create a new (empty) result message
294 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
295 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
296 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
297
298 String lang = request.getAttribute(GSXML.LANG_ATT);
299 // Get the parameters of the request
300 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
301 if (param_list == null) {
302 System.err.println("Error: DocumentMetadataRetrieve request had no paramList.");
303 return result; // Return the empty result
304 }
305
306 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
307 if (query_doc_list.getLength()==0) {
308 System.err.println("Error: DocumentMetadataRetrieve request had no documentNodes.");
309 return result; // Return the empty result
310 }
311
312 // the resulting doc node list
313 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
314 result.appendChild(result_node_list);
315
316 // get all the metadata params
317 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
318 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
319 new_param_list.appendChild(param);
320
321 // organise the nodes into collection lists
322 HashMap coll_map = new HashMap();
323
324 for (int i=0; i<query_doc_list.getLength(); i++) {
325 Element doc_node = (Element)query_doc_list.item(i);
326 String coll_name = doc_node.getAttribute("collection");
327 Element coll_items = (Element)coll_map.get(coll_name);
328 if (coll_items==null) {
329 coll_items=this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
330 coll_map.put(coll_name, coll_items);
331 }
332 coll_items.appendChild(this.doc.importNode(doc_node, true));
333 }
334
335 // create teh individual requests
336 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
337 Set mapping_set = coll_map.entrySet();
338 Iterator iter = mapping_set.iterator();
339
340 while (iter.hasNext()) {
341 Map.Entry e = (Map.Entry)iter.next();
342 String cname = (String)e.getKey();
343 Element doc_nodes = (Element)e.getValue();
344 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), lang, "");
345 meta_request.appendChild(doc_nodes);
346 meta_request.appendChild(new_param_list.cloneNode(true));
347 meta_request_message.appendChild(meta_request);
348
349 }
350 Element meta_result = this.router.process(meta_request_message);
351
352 // now need to put the doc nodes back in the right order
353 // go through the original list again. keep an element pointer to
354 // the next element in each collections list
355 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
356 for (int i=0; i<meta_responses.getLength(); i++) {
357 String collname = GSPath.getFirstLink(((Element)meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
358 Element first_elem = (Element)GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
359 coll_map.put(collname, first_elem);
360 }
361
362 for (int i=0; i<query_doc_list.getLength(); i++) {
363 Element doc_node = (Element)query_doc_list.item(i);
364 Element new_node = (Element)this.doc.importNode(doc_node, false);
365 result_node_list.appendChild(new_node);
366 String coll_name = doc_node.getAttribute("collection");
367
368 Element meta_elem = (Element)coll_map.get(coll_name);
369 GSXML.mergeMetadataLists(new_node, meta_elem);
370 coll_map.put(coll_name, meta_elem.getNextSibling());
371 }
372 return result;
373 }
374}
375
Note: See TracBrowser for help on using the repository browser.