source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 28180

Last change on this file since 28180 was 28180, checked in by kjdon, 11 years ago

removed my email address

  • Property svn:keywords set to Author Date Id Revision
File size: 19.4 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 */
38
39public class CrossCollectionSearch extends ServiceRack
40{
41
42 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.CrossCollectionSearch.class.getName());
43 protected static final String QUERY_PARAM = "query";
44 protected static final String COLLECTION_PARAM = "collection";
45
46 // the services on offer - these proxy the actual collection ones
47 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
48 protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
49
50 protected String[] coll_ids_list = null;
51 protected String[] coll_ids_list_no_all = null;
52 // maps lang to coll names list
53 protected HashMap<String, String[]> coll_names_map = null;
54
55 //protected String[] coll_names_list = null;
56
57 /** constructor */
58 public CrossCollectionSearch()
59 {
60 }
61
62 public boolean configure(Element info, Element extra_info)
63 {
64 // any parameters? colls to include??
65 logger.info("Configuring CrossCollectionSearch...");
66 // query service
67 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
68 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
69 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
70 this.short_service_info.appendChild(ccs_service);
71
72 // metadata service
73 Element dmr_service = this.doc.createElement(GSXML.SERVICE_ELEM);
74 dmr_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
75 dmr_service.setAttribute(GSXML.NAME_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
76 this.short_service_info.appendChild(dmr_service);
77
78 // get any format info
79 Element format_info = (Element) GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
80 if (format_info != null)
81 {
82 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format_info, true));
83 }
84 else
85 {
86 // add in a default format statement
87 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=";
88 if (this.cluster_name != null)
89 {
90 format_string += this.cluster_name;
91 }
92 format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
93 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
94 }
95 return true;
96 }
97
98 protected Element getServiceDescription(String service, String lang, String subset)
99 {
100 if (service.equals(TEXT_QUERY_SERVICE))
101 {
102
103 Element ccs_service = this.doc.createElement(GSXML.SERVICE_ELEM);
104 ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
105 ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
106
107 // display info
108 if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
109 {
110 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE + ".name", lang)));
111 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE + ".submit", lang)));
112 ccs_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE + ".description", lang)));
113 }
114 // param info
115 if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
116 {
117 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
118 // collection list
119 if (coll_ids_list == null)
120 {
121 initCollectionList(lang);
122 }
123 if (!coll_names_map.containsKey(lang))
124 {
125 addCollectionNames(lang);
126 }
127 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));
128 param_list.appendChild(param);
129 // query param
130 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param." + QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
131 param_list.appendChild(param);
132 ccs_service.appendChild(param_list);
133 }
134
135 logger.debug("service description=" + this.converter.getPrettyString(ccs_service));
136 return ccs_service;
137 }
138 // these ones are probably never called, but put them here just in case
139 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
140 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
141 service_elem.setAttribute(GSXML.NAME_ATT, service);
142 return service_elem;
143
144 }
145
146 protected Element processTextQuery(Element request)
147 {
148 // Create a new (empty) result message
149 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
150 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
151 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
152
153 UserContext userContext = new UserContext(request);
154
155 // Get the parameters of the request
156 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
157 if (param_list == null)
158 {
159 logger.error("TextQuery request had no paramList.");
160 return result; // Return the empty result
161 }
162
163 // get the collection list
164 String[] colls_list = coll_ids_list_no_all;
165 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
166 if (coll_param != null)
167 {
168 String coll_list = GSXML.getValue(coll_param);
169 if (!coll_list.equals("all") && !coll_list.equals(""))
170 {
171 colls_list = coll_list.split(",");
172 }
173 }
174
175 Element query_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
176 // we are sending the same request to each collection - build up the to
177 // attribute for the request
178 StringBuffer to_att = new StringBuffer();
179 for (int i = 0; i < colls_list.length; i++)
180 {
181 if (i > 0)
182 {
183 to_att.append(",");
184 }
185 to_att.append(GSPath.appendLink(colls_list[i], "TextQuery"));
186
187 }
188 // send the query to all colls
189 Element query_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), userContext);
190 query_message.appendChild(query_request);
191 // should we add params individually?
192 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
193 query_request.appendChild(new_param_list);
194 new_param_list.appendChild(this.doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
195 Element query_result = (Element) this.router.process(query_message);
196
197 // gather up the data from each response
198 int numDocsMatched = 0;
199 int numDocsReturned = 0;
200
201 //term info??
202
203 NodeList metadata = query_result.getElementsByTagName(GSXML.METADATA_ELEM);
204 for (int j = 0; j < metadata.getLength(); j++)
205 {
206 Element meta = (Element) metadata.item(j);
207 if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsReturned"))
208 {
209 numDocsReturned += Integer.parseInt(GSXML.getValue(meta));
210 }
211 else if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsMatched"))
212 {
213 numDocsMatched += Integer.parseInt(GSXML.getValue(meta));
214 }
215 }
216
217 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
218 result.appendChild(metadata_list);
219 GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", "" + numDocsReturned);
220 //GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numDocsMatched);
221
222 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
223 result.appendChild(doc_node_list);
224
225 NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
226
227 for (int k = 0; k < responses.getLength(); k++)
228 {
229 String coll_name = GSPath.removeLastLink(((Element) responses.item(k)).getAttribute(GSXML.FROM_ATT));
230 NodeList nodes = ((Element) responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
231 if (nodes == null || nodes.getLength() == 0)
232 continue;
233 Element last_node = null;
234 Element this_node = null;
235 for (int n = 0; n < nodes.getLength(); n++)
236 {
237 this_node = (Element) nodes.item(n);
238 this_node.setAttribute("collection", coll_name);
239 if (k == 0)
240 {
241
242 doc_node_list.appendChild(this.doc.importNode(this_node, true));
243 }
244 else
245 {
246 if (last_node == null)
247 {
248 last_node = (Element) GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
249 }
250 last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
251 }
252
253 }
254 }
255 return result;
256 }
257
258 // protected Element processAdvTextQuery(Element request)
259 // {
260
261 // }
262 protected boolean initCollectionList(String lang)
263 {
264 UserContext userContext = new UserContext();
265 userContext.setLanguage(lang);
266 userContext.setUserID("");
267
268 // first, get the message router info
269 Element coll_list_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
270 Element coll_list_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, "", userContext); // uid
271 coll_list_message.appendChild(coll_list_request);
272 logger.debug("coll list request = " + this.converter.getPrettyString(coll_list_request));
273 Element coll_list_response = (Element) this.router.process(coll_list_message);
274 if (coll_list_response == null)
275 {
276 logger.error("couldn't query the message router!");
277 return false;
278 }
279 logger.debug("coll list response = " + this.converter.getPrettyString(coll_list_response));
280 // second, get some info from each collection. we want the coll name
281 // and whether its got a text query service
282
283 NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
284 // we can send the same request to multiple collections at once by using a comma separated list
285 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
286 StringBuffer colls_sb = new StringBuffer();
287 for (int i = 0; i < colls.getLength(); i++)
288 {
289 Element c = (Element) colls.item(i);
290 String name = c.getAttribute(GSXML.NAME_ATT);
291 if (i != 0)
292 {
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
300 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
301 metadata_message.appendChild(metadata_request);
302 logger.debug("metadata request = " + this.converter.getPrettyString(metadata_message));
303 Element metadata_response = (Element) this.router.process(metadata_message);
304 logger.debug("metadata response = " + this.converter.getPrettyString(metadata_response));
305 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
306 ArrayList<String> valid_colls = new ArrayList<String>();
307 ArrayList<String> valid_coll_names = new ArrayList<String>();
308 for (int i = 0; i < coll_responses.getLength(); i++)
309 {
310 Element response = (Element) coll_responses.item(i);
311 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
312 Element service_list = (Element) GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
313 if (service_list == null)
314 continue;
315 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
316 if (query_service == null)
317 continue;
318 // use the name of the response in case we are talking to a remote collection, not the name of the collection.
319 String coll_id = response.getAttribute(GSXML.FROM_ATT);
320 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
321 valid_colls.add(coll_id);
322 valid_coll_names.add(coll_name);
323 }
324
325 this.coll_names_map = new HashMap<String, String[]>();
326
327 // ids no all has the list without 'all' option.
328 this.coll_ids_list_no_all = new String[1];
329 this.coll_ids_list_no_all = valid_colls.toArray(coll_ids_list_no_all);
330
331 valid_colls.add(0, "all");
332 valid_coll_names.add(0, getTextString("param." + COLLECTION_PARAM + ".all", lang));
333
334 this.coll_ids_list = new String[1];
335 this.coll_ids_list = valid_colls.toArray(coll_ids_list);
336
337 String[] coll_names_list = new String[1];
338 coll_names_list = valid_coll_names.toArray(coll_names_list);
339 this.coll_names_map.put(lang, coll_names_list);
340 return true;
341 }
342
343 protected void addCollectionNames(String lang)
344 {
345
346 UserContext userContext = new UserContext();
347 userContext.setLanguage(lang);
348 userContext.setUserID("");
349
350 ArrayList<String> coll_names = new ArrayList<String>();
351 coll_names.add(getTextString("param." + COLLECTION_PARAM + ".all", lang));
352
353 // need to request MR for collection descriptions
354 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
355
356 // get a comma separated list of coll ids to send to MR
357 // the first item is the place holder for 'all'
358 StringBuffer colls_sb = new StringBuffer();
359 for (int i = 1; i < coll_ids_list.length; i++)
360 {
361 if (i != 1)
362 {
363 colls_sb.append(",");
364 }
365 colls_sb.append(coll_ids_list[i]);
366 }
367 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
368 // param_list to request just displayTextList
369 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
370 Element param = GSXML.createParameter(this.doc, GSXML.SUBSET_PARAM, GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER);
371 param_list.appendChild(param);
372 metadata_request.appendChild(param_list);
373 metadata_message.appendChild(metadata_request);
374 logger.debug("coll names metadata request = " + this.converter.getPrettyString(metadata_message));
375 Element metadata_response = (Element) this.router.process(metadata_message);
376 logger.debug("coll names metadata response = " + this.converter.getPrettyString(metadata_response));
377 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
378 for (int i = 0; i < coll_responses.getLength(); i++)
379 {
380 Element response = (Element) coll_responses.item(i);
381 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
382 String coll_name = GSXML.getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
383 coll_names.add(coll_name);
384 }
385
386 String[] coll_names_list = new String[1];
387 coll_names_list = coll_names.toArray(coll_names_list);
388 this.coll_names_map.put(lang, coll_names_list);
389
390 }
391
392 protected Element processDocumentMetadataRetrieve(Element request)
393 {
394 // Create a new (empty) result message
395 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
396 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
397 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
398
399 UserContext userContext = new UserContext(request);
400 // Get the parameters of the request
401 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
402 if (param_list == null)
403 {
404 logger.error("DocumentMetadataRetrieve request had no paramList.");
405 return result; // Return the empty result
406 }
407
408 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
409 if (query_doc_list.getLength() == 0)
410 {
411 logger.error("DocumentMetadataRetrieve request had no documentNodes.");
412 return result; // Return the empty result
413 }
414
415 // the resulting doc node list
416 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
417 result.appendChild(result_node_list);
418
419 // get all the metadata params
420 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
421 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
422 new_param_list.appendChild(param);
423
424 // organise the nodes into collection lists
425 HashMap<String, Node> coll_map = new HashMap<String, Node>();
426
427 for (int i = 0; i < query_doc_list.getLength(); i++)
428 {
429 Element doc_node = (Element) query_doc_list.item(i);
430 String coll_name = doc_node.getAttribute("collection");
431 Element coll_items = (Element) coll_map.get(coll_name);
432 if (coll_items == null)
433 {
434 coll_items = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
435 coll_map.put(coll_name, coll_items);
436 }
437 coll_items.appendChild(this.doc.importNode(doc_node, true));
438 }
439
440 // create teh individual requests
441 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
442 Set mapping_set = coll_map.entrySet();
443 Iterator iter = mapping_set.iterator();
444
445 while (iter.hasNext())
446 {
447 Map.Entry e = (Map.Entry) iter.next();
448 String cname = (String) e.getKey();
449 Element doc_nodes = (Element) e.getValue();
450 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), userContext);
451 meta_request.appendChild(doc_nodes);
452 meta_request.appendChild(new_param_list.cloneNode(true));
453 meta_request_message.appendChild(meta_request);
454
455 }
456
457 Node meta_result_node = this.router.process(meta_request_message);
458 Element meta_result = this.converter.nodeToElement(meta_result_node);
459
460 // now need to put the doc nodes back in the right order
461 // go through the original list again. keep an element pointer to
462 // the next element in each collections list
463 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
464 for (int i = 0; i < meta_responses.getLength(); i++)
465 {
466 String collname = GSPath.removeLastLink(((Element) meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
467 Element first_elem = (Element) GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
468 coll_map.put(collname, first_elem);
469 }
470
471 for (int i = 0; i < query_doc_list.getLength(); i++)
472 {
473 Element doc_node = (Element) query_doc_list.item(i);
474 Element new_node = (Element) this.doc.importNode(doc_node, false);
475 result_node_list.appendChild(new_node);
476 String coll_name = doc_node.getAttribute("collection");
477
478 Element meta_elem = (Element) coll_map.get(coll_name);
479 GSXML.mergeMetadataLists(new_node, meta_elem);
480 coll_map.put(coll_name, meta_elem.getNextSibling());
481 }
482 return result;
483 }
484}
Note: See TracBrowser for help on using the repository browser.