source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 26406

Last change on this file since 26406 was 26406, checked in by kjdon, 11 years ago

made the list of collections use language dependent descriptions. added a hashmap to store descriptions by language

  • Property svn:keywords set to Author Date Id Revision
File size: 19.6 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;
23import org.greenstone.gsdl3.util.UserContext;
24
25import org.w3c.dom.Element;
26import org.w3c.dom.Node;
27import org.w3c.dom.NodeList;
28
29import java.util.ArrayList;
30import java.util.Map;
31import java.util.Iterator;
32import java.util.Set;
33import java.util.HashMap;
34
35import org.apache.log4j.*;
36
37/**
38 *
39 * @author <a href="mailto:[email protected]">Katherine Don</a>
40 */
41
42public class CrossCollectionSearch extends ServiceRack
43{
44
45 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.CrossCollectionSearch.class.getName());
46 protected static final String QUERY_PARAM = "query";
47 protected static final String COLLECTION_PARAM = "collection";
48
49 // the services on offer - these proxy the actual collection ones
50 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
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 // maps lang to coll names list
56 protected HashMap<String, String[]> coll_names_map = null;
57 //protected String[] coll_names_list = null;
58
59 /** constructor */
60 public CrossCollectionSearch()
61 {
62 }
63
64 public boolean configure(Element info, Element extra_info)
65 {
66 // any parameters? colls to include??
67 logger.info("Configuring CrossCollectionSearch...");
68 // query service
69 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
70 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
71 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
72 this.short_service_info.appendChild(ccs_service);
73
74 // metadata service
75 Element dmr_service = this.doc.createElement(GSXML.SERVICE_ELEM);
76 dmr_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
77 dmr_service.setAttribute(GSXML.NAME_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
78 this.short_service_info.appendChild(dmr_service);
79
80 // get any format info
81 Element format_info = (Element) GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
82 if (format_info != null)
83 {
84 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format_info, true));
85 }
86 else
87 {
88 // add in a default format statement
89 String format_string = "<format xmlns:gsf='" + GSXML.GSF_NAMESPACE + "' xmlns:xsl='" + GSXML.XSL_NAMESPACE + "'><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=";
90 if (this.cluster_name != null)
91 {
92 format_string += this.cluster_name;
93 }
94 format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
95 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
96 }
97 return true;
98 }
99
100 protected Element getServiceDescription(String service, String lang, String subset)
101 {
102 if (service.equals(TEXT_QUERY_SERVICE))
103 {
104
105 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
106 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
107 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
108
109 // display info
110 if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
111 {
112 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE + ".name", lang)));
113 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE + ".submit", lang)));
114 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE + ".description", lang)));
115 }
116 // param info
117 if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
118 {
119 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
120 // collection list
121 if (coll_ids_list == null)
122 {
123 initCollectionList(lang);
124 }
125 if (!coll_names_map.containsKey(lang)) {
126 addCollectionNames(lang);
127 }
128 Element param = GSXML.createParameterDescription(this.doc, COLLECTION_PARAM, getTextString("param." + COLLECTION_PARAM, lang), GSXML.PARAM_TYPE_ENUM_MULTI, "all", coll_ids_list, coll_names_map.get(lang));
129 param_list.appendChild(param);
130 // query param
131 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param." + QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
132 param_list.appendChild(param);
133 ccs_service.appendChild(param_list);
134 }
135
136 logger.debug("service description=" + this.converter.getPrettyString(ccs_service));
137 return ccs_service;
138 }
139 // these ones are probably never called, but put them here just in case
140 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
141 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
142 service_elem.setAttribute(GSXML.NAME_ATT, service);
143 return service_elem;
144
145 }
146
147 protected Element processTextQuery(Element request)
148 {
149 // Create a new (empty) result message
150 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
151 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
152 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
153
154 UserContext userContext = new UserContext(request);
155
156 // Get the parameters of the request
157 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
158 if (param_list == null)
159 {
160 logger.error("TextQuery request had no paramList.");
161 return result; // Return the empty result
162 }
163
164 // get the collection list
165 String[] colls_list = coll_ids_list_no_all;
166 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
167 if (coll_param != null)
168 {
169 String coll_list = GSXML.getValue(coll_param);
170 if (!coll_list.equals("all") && !coll_list.equals(""))
171 {
172 colls_list = coll_list.split(",");
173 }
174 }
175
176 Element query_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
177 // we are sending the same request to each collection - build up the to
178 // attribute for the request
179 StringBuffer to_att = new StringBuffer();
180 for (int i = 0; i < colls_list.length; i++)
181 {
182 if (i > 0)
183 {
184 to_att.append(",");
185 }
186 to_att.append(GSPath.appendLink(colls_list[i], "TextQuery"));
187
188 }
189 // send the query to all colls
190 Element query_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), userContext);
191 query_message.appendChild(query_request);
192 // should we add params individually?
193 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
194 query_request.appendChild(new_param_list);
195 new_param_list.appendChild(this.doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
196 Element query_result = (Element) this.router.process(query_message);
197
198 // gather up the data from each response
199 int numDocsMatched = 0;
200 int numDocsReturned = 0;
201
202 //term info??
203
204 NodeList metadata = query_result.getElementsByTagName(GSXML.METADATA_ELEM);
205 for (int j = 0; j < metadata.getLength(); j++)
206 {
207 Element meta = (Element) metadata.item(j);
208 if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsReturned"))
209 {
210 numDocsReturned += Integer.parseInt(GSXML.getValue(meta));
211 }
212 else if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsMatched"))
213 {
214 numDocsMatched += Integer.parseInt(GSXML.getValue(meta));
215 }
216 }
217
218 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
219 result.appendChild(metadata_list);
220 GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", "" + numDocsReturned);
221 //GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numDocsMatched);
222
223 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
224 result.appendChild(doc_node_list);
225
226 NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
227
228 for (int k = 0; k < responses.getLength(); k++)
229 {
230 String coll_name = GSPath.removeLastLink(((Element) responses.item(k)).getAttribute(GSXML.FROM_ATT));
231 NodeList nodes = ((Element) responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
232 if (nodes == null || nodes.getLength() == 0)
233 continue;
234 Element last_node = null;
235 Element this_node = null;
236 for (int n = 0; n < nodes.getLength(); n++)
237 {
238 this_node = (Element) nodes.item(n);
239 this_node.setAttribute("collection", coll_name);
240 if (k == 0)
241 {
242
243 doc_node_list.appendChild(this.doc.importNode(this_node, true));
244 }
245 else
246 {
247 if (last_node == null)
248 {
249 last_node = (Element) GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
250 }
251 last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
252 }
253
254 }
255 }
256 return result;
257 }
258
259 // protected Element processAdvTextQuery(Element request)
260 // {
261
262 // }
263 protected boolean initCollectionList(String lang)
264 {
265 UserContext userContext = new UserContext();
266 userContext.setLanguage(lang);
267 userContext.setUserID("");
268
269 // first, get the message router info
270 Element coll_list_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
271 Element coll_list_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, "", userContext); // uid
272 coll_list_message.appendChild(coll_list_request);
273 logger.debug("coll list request = " + this.converter.getPrettyString(coll_list_request));
274 Element coll_list_response = (Element) this.router.process(coll_list_message);
275 if (coll_list_response == null)
276 {
277 logger.error("couldn't query the message router!");
278 return false;
279 }
280 logger.debug("coll list response = " + this.converter.getPrettyString(coll_list_response));
281 // second, get some info from each collection. we want the coll name
282 // and whether its got a text query service
283
284 NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
285 // we can send the same request to multiple collections at once by using a comma separated list
286 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
287 StringBuffer colls_sb = new StringBuffer();
288 for (int i = 0; i < colls.getLength(); i++)
289 {
290 Element c = (Element) colls.item(i);
291 String name = c.getAttribute(GSXML.NAME_ATT);
292 if (i!=0) {
293 colls_sb.append(",");
294 }
295 colls_sb.append(name);
296 //Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, userContext);
297 //metadata_message.appendChild(metadata_request);
298 }
299 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
300 metadata_message.appendChild(metadata_request);
301 logger.debug("metadata request = " + this.converter.getPrettyString(metadata_message));
302 Element metadata_response = (Element) this.router.process(metadata_message);
303 logger.debug("metadata response = " + this.converter.getPrettyString(metadata_response));
304 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
305 ArrayList<String> valid_colls = new ArrayList<String>();
306 ArrayList<String> valid_coll_names = new ArrayList<String>();
307 for (int i = 0; i < coll_responses.getLength(); i++)
308 {
309 Element response = (Element) coll_responses.item(i);
310 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
311 Element service_list = (Element) GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
312 if (service_list == null)
313 continue;
314 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
315 if (query_service == null)
316 continue;
317 // use the name of the response in case we are talking to a remote collection, not the name of the collection.
318 String coll_id = response.getAttribute(GSXML.FROM_ATT);
319 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
320 valid_colls.add(coll_id);
321 valid_coll_names.add(coll_name);
322 }
323
324 if (valid_colls.size() == 0)
325 {
326 return false;
327 }
328 // ids no all has the list without 'all' option.
329 this.coll_ids_list_no_all = new String[1];
330 this.coll_ids_list_no_all = valid_colls.toArray(coll_ids_list_no_all);
331
332 valid_colls.add(0, "all");
333 valid_coll_names.add(0, getTextString("param." + COLLECTION_PARAM + ".all", lang));
334
335 this.coll_ids_list = new String[1];
336 this.coll_ids_list = valid_colls.toArray(coll_ids_list);
337
338 this.coll_names_map = new HashMap<String, String[]>();
339 String[] coll_names_list = new String[1];
340 coll_names_list = valid_coll_names.toArray(coll_names_list);
341 this.coll_names_map.put(lang, coll_names_list);
342 return true;
343 }
344
345 protected void addCollectionNames(String lang) {
346
347 UserContext userContext = new UserContext();
348 userContext.setLanguage(lang);
349 userContext.setUserID("");
350
351 ArrayList<String> coll_names = new ArrayList<String>();
352 coll_names.add(getTextString("param." + COLLECTION_PARAM + ".all", lang));
353
354 // need to request MR for collection descriptions
355 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
356
357 // get a comma separated list of coll ids to send to MR
358 // the first item is the place holder for 'all'
359 StringBuffer colls_sb = new StringBuffer();
360 for (int i=1; i<coll_ids_list.length; i++) {
361 if (i!=1) {
362 colls_sb.append(",");
363 }
364 colls_sb.append(coll_ids_list[i]);
365 }
366 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
367 // param_list to request just displayTextList
368 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
369 Element param = GSXML.createParameter(this.doc, GSXML.SUBSET_PARAM, GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER);
370 param_list.appendChild(param);
371 metadata_request.appendChild(param_list);
372 metadata_message.appendChild(metadata_request);
373 logger.debug("coll names metadata request = " + this.converter.getPrettyString(metadata_message));
374 Element metadata_response = (Element) this.router.process(metadata_message);
375 logger.debug("coll names metadata response = " + this.converter.getPrettyString(metadata_response));
376 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
377 for (int i = 0; i < coll_responses.getLength(); i++)
378 {
379 Element response = (Element) coll_responses.item(i);
380 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
381 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
382 coll_names.add(coll_name);
383 }
384
385 String[] coll_names_list = new String[1];
386 coll_names_list = coll_names.toArray(coll_names_list);
387 this.coll_names_map.put(lang, coll_names_list);
388
389 }
390
391 protected Element processDocumentMetadataRetrieve(Element request)
392 {
393 // Create a new (empty) result message
394 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
395 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
396 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
397
398 UserContext userContext = new UserContext(request);
399 // Get the parameters of the request
400 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
401 if (param_list == null)
402 {
403 logger.error("DocumentMetadataRetrieve request had no paramList.");
404 return result; // Return the empty result
405 }
406
407 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
408 if (query_doc_list.getLength() == 0)
409 {
410 logger.error("DocumentMetadataRetrieve request had no documentNodes.");
411 return result; // Return the empty result
412 }
413
414 // the resulting doc node list
415 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
416 result.appendChild(result_node_list);
417
418 // get all the metadata params
419 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
420 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
421 new_param_list.appendChild(param);
422
423 // organise the nodes into collection lists
424 HashMap<String, Node> coll_map = new HashMap<String, Node>();
425
426 for (int i = 0; i < query_doc_list.getLength(); i++)
427 {
428 Element doc_node = (Element) query_doc_list.item(i);
429 String coll_name = doc_node.getAttribute("collection");
430 Element coll_items = (Element) coll_map.get(coll_name);
431 if (coll_items == null)
432 {
433 coll_items = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
434 coll_map.put(coll_name, coll_items);
435 }
436 coll_items.appendChild(this.doc.importNode(doc_node, true));
437 }
438
439 // create teh individual requests
440 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
441 Set mapping_set = coll_map.entrySet();
442 Iterator iter = mapping_set.iterator();
443
444 while (iter.hasNext())
445 {
446 Map.Entry e = (Map.Entry) iter.next();
447 String cname = (String) e.getKey();
448 Element doc_nodes = (Element) e.getValue();
449 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), userContext);
450 meta_request.appendChild(doc_nodes);
451 meta_request.appendChild(new_param_list.cloneNode(true));
452 meta_request_message.appendChild(meta_request);
453
454 }
455
456 Node meta_result_node = this.router.process(meta_request_message);
457 Element meta_result = this.converter.nodeToElement(meta_result_node);
458
459 // now need to put the doc nodes back in the right order
460 // go through the original list again. keep an element pointer to
461 // the next element in each collections list
462 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
463 for (int i = 0; i < meta_responses.getLength(); i++)
464 {
465 String collname = GSPath.removeLastLink(((Element) meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
466 Element first_elem = (Element) GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
467 coll_map.put(collname, first_elem);
468 }
469
470 for (int i = 0; i < query_doc_list.getLength(); i++)
471 {
472 Element doc_node = (Element) query_doc_list.item(i);
473 Element new_node = (Element) this.doc.importNode(doc_node, false);
474 result_node_list.appendChild(new_node);
475 String coll_name = doc_node.getAttribute("collection");
476
477 Element meta_elem = (Element) coll_map.get(coll_name);
478 GSXML.mergeMetadataLists(new_node, meta_elem);
479 coll_map.put(coll_name, meta_elem.getNextSibling());
480 }
481 return result;
482 }
483}
Note: See TracBrowser for help on using the repository browser.