source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 26578

Last change on this file since 26578 was 26578, checked in by sjm84, 11 years ago

Removing a test that was causing a null pointer exception

  • Property svn:keywords set to Author Date Id Revision
File size: 19.5 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 java.util.ArrayList;
22import java.util.HashMap;
23import java.util.Iterator;
24import java.util.Map;
25import java.util.Set;
26
27import org.apache.log4j.Logger;
28import org.greenstone.gsdl3.util.GSPath;
29import org.greenstone.gsdl3.util.GSXML;
30import org.greenstone.gsdl3.util.UserContext;
31import org.w3c.dom.Element;
32import org.w3c.dom.Node;
33import org.w3c.dom.NodeList;
34
35/**
36 *
37 * @author <a href="mailto:[email protected]">Katherine Don</a>
38 */
39
40public class CrossCollectionSearch extends ServiceRack
41{
42
43 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.CrossCollectionSearch.class.getName());
44 protected static final String QUERY_PARAM = "query";
45 protected static final String COLLECTION_PARAM = "collection";
46
47 // the services on offer - these proxy the actual collection ones
48 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
49 protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
50
51 protected String[] coll_ids_list = null;
52 protected String[] coll_ids_list_no_all = null;
53 // maps lang to coll names list
54 protected HashMap<String, String[]> coll_names_map = null;
55
56 //protected String[] coll_names_list = null;
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 {
83 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format_info, true));
84 }
85 else
86 {
87 // add in a default format statement
88 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=";
89 if (this.cluster_name != null)
90 {
91 format_string += this.cluster_name;
92 }
93 format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
94 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
95 }
96 return true;
97 }
98
99 protected Element getServiceDescription(String service, String lang, String subset)
100 {
101 if (service.equals(TEXT_QUERY_SERVICE))
102 {
103
104 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
105 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
106 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
107
108 // display info
109 if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
110 {
111 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE + ".name", lang)));
112 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE + ".submit", lang)));
113 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE + ".description", lang)));
114 }
115 // param info
116 if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
117 {
118 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
119 // collection list
120 if (coll_ids_list == null)
121 {
122 initCollectionList(lang);
123 }
124 if (!coll_names_map.containsKey(lang))
125 {
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 {
294 colls_sb.append(",");
295 }
296 colls_sb.append(name);
297 //Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, userContext);
298 //metadata_message.appendChild(metadata_request);
299 }
300
301 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
302 metadata_message.appendChild(metadata_request);
303 logger.debug("metadata request = " + this.converter.getPrettyString(metadata_message));
304 Element metadata_response = (Element) this.router.process(metadata_message);
305 logger.debug("metadata response = " + this.converter.getPrettyString(metadata_response));
306 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
307 ArrayList<String> valid_colls = new ArrayList<String>();
308 ArrayList<String> valid_coll_names = new ArrayList<String>();
309 for (int i = 0; i < coll_responses.getLength(); i++)
310 {
311 Element response = (Element) coll_responses.item(i);
312 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
313 Element service_list = (Element) GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
314 if (service_list == null)
315 continue;
316 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
317 if (query_service == null)
318 continue;
319 // use the name of the response in case we are talking to a remote collection, not the name of the collection.
320 String coll_id = response.getAttribute(GSXML.FROM_ATT);
321 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
322 valid_colls.add(coll_id);
323 valid_coll_names.add(coll_name);
324 }
325
326 this.coll_names_map = new HashMap<String, String[]>();
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 String[] coll_names_list = new String[1];
339 coll_names_list = valid_coll_names.toArray(coll_names_list);
340 this.coll_names_map.put(lang, coll_names_list);
341 return true;
342 }
343
344 protected void addCollectionNames(String lang)
345 {
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 {
362 if (i != 1)
363 {
364 colls_sb.append(",");
365 }
366 colls_sb.append(coll_ids_list[i]);
367 }
368 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
369 // param_list to request just displayTextList
370 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
371 Element param = GSXML.createParameter(this.doc, GSXML.SUBSET_PARAM, GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER);
372 param_list.appendChild(param);
373 metadata_request.appendChild(param_list);
374 metadata_message.appendChild(metadata_request);
375 logger.debug("coll names metadata request = " + this.converter.getPrettyString(metadata_message));
376 Element metadata_response = (Element) this.router.process(metadata_message);
377 logger.debug("coll names metadata response = " + this.converter.getPrettyString(metadata_response));
378 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
379 for (int i = 0; i < coll_responses.getLength(); i++)
380 {
381 Element response = (Element) coll_responses.item(i);
382 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
383 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
384 coll_names.add(coll_name);
385 }
386
387 String[] coll_names_list = new String[1];
388 coll_names_list = coll_names.toArray(coll_names_list);
389 this.coll_names_map.put(lang, coll_names_list);
390
391 }
392
393 protected Element processDocumentMetadataRetrieve(Element request)
394 {
395 // Create a new (empty) result message
396 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
397 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
398 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
399
400 UserContext userContext = new UserContext(request);
401 // Get the parameters of the request
402 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
403 if (param_list == null)
404 {
405 logger.error("DocumentMetadataRetrieve request had no paramList.");
406 return result; // Return the empty result
407 }
408
409 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
410 if (query_doc_list.getLength() == 0)
411 {
412 logger.error("DocumentMetadataRetrieve request had no documentNodes.");
413 return result; // Return the empty result
414 }
415
416 // the resulting doc node list
417 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
418 result.appendChild(result_node_list);
419
420 // get all the metadata params
421 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
422 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
423 new_param_list.appendChild(param);
424
425 // organise the nodes into collection lists
426 HashMap<String, Node> coll_map = new HashMap<String, Node>();
427
428 for (int i = 0; i < query_doc_list.getLength(); i++)
429 {
430 Element doc_node = (Element) query_doc_list.item(i);
431 String coll_name = doc_node.getAttribute("collection");
432 Element coll_items = (Element) coll_map.get(coll_name);
433 if (coll_items == null)
434 {
435 coll_items = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
436 coll_map.put(coll_name, coll_items);
437 }
438 coll_items.appendChild(this.doc.importNode(doc_node, true));
439 }
440
441 // create teh individual requests
442 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
443 Set mapping_set = coll_map.entrySet();
444 Iterator iter = mapping_set.iterator();
445
446 while (iter.hasNext())
447 {
448 Map.Entry e = (Map.Entry) iter.next();
449 String cname = (String) e.getKey();
450 Element doc_nodes = (Element) e.getValue();
451 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), userContext);
452 meta_request.appendChild(doc_nodes);
453 meta_request.appendChild(new_param_list.cloneNode(true));
454 meta_request_message.appendChild(meta_request);
455
456 }
457
458 Node meta_result_node = this.router.process(meta_request_message);
459 Element meta_result = this.converter.nodeToElement(meta_result_node);
460
461 // now need to put the doc nodes back in the right order
462 // go through the original list again. keep an element pointer to
463 // the next element in each collections list
464 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
465 for (int i = 0; i < meta_responses.getLength(); i++)
466 {
467 String collname = GSPath.removeLastLink(((Element) meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
468 Element first_elem = (Element) GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
469 coll_map.put(collname, first_elem);
470 }
471
472 for (int i = 0; i < query_doc_list.getLength(); i++)
473 {
474 Element doc_node = (Element) query_doc_list.item(i);
475 Element new_node = (Element) this.doc.importNode(doc_node, false);
476 result_node_list.appendChild(new_node);
477 String coll_name = doc_node.getAttribute("collection");
478
479 Element meta_elem = (Element) coll_map.get(coll_name);
480 GSXML.mergeMetadataLists(new_node, meta_elem);
481 coll_map.put(coll_name, meta_elem.getNextSibling());
482 }
483 return result;
484 }
485}
Note: See TracBrowser for help on using the repository browser.