source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/CrossCollectionSearch.java@ 26055

Last change on this file since 26055 was 26055, checked in by sjm84, 12 years ago

Adding constants for the various XML namespaces used in the code
  • Property svn:keywords set to Author Date Id Revision
File size: 17.0 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 ADV_QUERY_SERVICE = "AdvTextQuery";
52 protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
53
54 protected String[] coll_ids_list = null;
55 protected String[] coll_ids_list_no_all = null;
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();
123 }
124 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);
125 param_list.appendChild(param);
126 // query param
127 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param." + QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
128 param_list.appendChild(param);
129 ccs_service.appendChild(param_list);
130 }
131
132 logger.debug("service description=" + this.converter.getPrettyString(ccs_service));
133 return ccs_service;
134 }
135 // these ones are probably never called, but put them here just in case
136 Element service_elem = this.doc.createElement(GSXML.SERVICE_ELEM);
137 service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
138 service_elem.setAttribute(GSXML.NAME_ATT, service);
139 return service_elem;
140
141 }
142
143 protected Element processTextQuery(Element request)
144 {
145 // Create a new (empty) result message
146 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
147 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
148 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
149
150 UserContext userContext = new UserContext(request);
151
152 // Get the parameters of the request
153 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
154 if (param_list == null)
155 {
156 logger.error("TextQuery request had no paramList.");
157 return result; // Return the empty result
158 }
159
160 // get the collection list
161 String[] colls_list = coll_ids_list_no_all;
162 Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
163 if (coll_param != null)
164 {
165 String coll_list = GSXML.getValue(coll_param);
166 if (!coll_list.equals("all") && !coll_list.equals(""))
167 {
168 colls_list = coll_list.split(",");
169 }
170 }
171
172 Element query_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
173 // we are sending the same request to each collection - build up the to
174 // attribute for the request
175 StringBuffer to_att = new StringBuffer();
176 for (int i = 0; i < colls_list.length; i++)
177 {
178 if (i > 0)
179 {
180 to_att.append(",");
181 }
182 to_att.append(GSPath.appendLink(colls_list[i], "TextQuery"));
183
184 }
185 // send the query to all colls
186 Element query_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), userContext);
187 query_message.appendChild(query_request);
188 // should we add params individually?
189 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
190 query_request.appendChild(new_param_list);
191 new_param_list.appendChild(this.doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
192 Element query_result = (Element) this.router.process(query_message);
193
194 // gather up the data from each response
195 int numDocsMatched = 0;
196 int numDocsReturned = 0;
197
198 //term info??
199
200 NodeList metadata = query_result.getElementsByTagName(GSXML.METADATA_ELEM);
201 for (int j = 0; j < metadata.getLength(); j++)
202 {
203 Element meta = (Element) metadata.item(j);
204 if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsReturned"))
205 {
206 numDocsReturned += Integer.parseInt(GSXML.getValue(meta));
207 }
208 else if (meta.getAttribute(GSXML.NAME_ATT).equals("numDocsMatched"))
209 {
210 numDocsMatched += Integer.parseInt(GSXML.getValue(meta));
211 }
212 }
213
214 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
215 result.appendChild(metadata_list);
216 GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", "" + numDocsReturned);
217 //GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numDocsMatched);
218
219 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
220 result.appendChild(doc_node_list);
221
222 NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
223
224 for (int k = 0; k < responses.getLength(); k++)
225 {
226 String coll_name = GSPath.removeLastLink(((Element) responses.item(k)).getAttribute(GSXML.FROM_ATT));
227 NodeList nodes = ((Element) responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
228 if (nodes == null || nodes.getLength() == 0)
229 continue;
230 Element last_node = null;
231 Element this_node = null;
232 for (int n = 0; n < nodes.getLength(); n++)
233 {
234 this_node = (Element) nodes.item(n);
235 this_node.setAttribute("collection", coll_name);
236 if (k == 0)
237 {
238
239 doc_node_list.appendChild(this.doc.importNode(this_node, true));
240 }
241 else
242 {
243 if (last_node == null)
244 {
245 last_node = (Element) GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
246 }
247 last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
248 }
249
250 }
251 }
252 return result;
253 }
254
255 // protected Element processAdvTextQuery(Element request)
256 // {
257
258 // }
259 protected boolean initCollectionList()
260 {
261 UserContext userContext = new UserContext();
262 userContext.setLanguage("en");
263 userContext.setUserID("");
264
265 // first, get the message router info
266 Element coll_list_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
267 Element coll_list_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, "", userContext); // uid
268 coll_list_message.appendChild(coll_list_request);
269 logger.debug("coll list request = " + this.converter.getPrettyString(coll_list_request));
270 Element coll_list_response = (Element) this.router.process(coll_list_message);
271 if (coll_list_response == null)
272 {
273 logger.error("couldn't query the message router!");
274 return false;
275 }
276 logger.debug("coll list response = " + this.converter.getPrettyString(coll_list_response));
277 // second, get some info from each collection. we want the coll name
278 // and whether its got a text query service
279
280 NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
281 // we will send all the requests in a single message
282 Element metadata_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
283 for (int i = 0; i < colls.getLength(); i++)
284 {
285 Element c = (Element) colls.item(i);
286 String name = c.getAttribute(GSXML.NAME_ATT);
287 Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, userContext);
288 metadata_message.appendChild(metadata_request);
289 }
290 logger.debug("metadata request = " + this.converter.getPrettyString(metadata_message));
291 Element metadata_response = (Element) this.router.process(metadata_message);
292 logger.debug("metadata response = " + this.converter.getPrettyString(metadata_response));
293 NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
294 ArrayList<String> valid_colls = new ArrayList<String>();
295 ArrayList<String> valid_coll_names = new ArrayList<String>();
296 for (int i = 0; i < coll_responses.getLength(); i++)
297 {
298 Element response = (Element) coll_responses.item(i);
299 Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
300 Element service_list = (Element) GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
301 if (service_list == null)
302 continue;
303 Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
304 if (query_service == null)
305 continue;
306 // use the name of the response in case we are talking to a remote collection, not the name of the collection.
307 String coll_id = response.getAttribute(GSXML.FROM_ATT);
308 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
309 valid_colls.add(coll_id);
310 valid_coll_names.add(coll_name);
311 }
312
313 if (valid_colls.size() == 0)
314 {
315 return false;
316 }
317 // ids no all has the list without 'all' option.
318 this.coll_ids_list_no_all = new String[1];
319 this.coll_ids_list_no_all = valid_colls.toArray(coll_ids_list_no_all);
320
321 valid_colls.add(0, "all");
322 valid_coll_names.add(0, getTextString("param." + COLLECTION_PARAM + ".all", "en"));
323 this.coll_ids_list = new String[1];
324 this.coll_names_list = new String[1];
325 this.coll_ids_list = valid_colls.toArray(coll_ids_list);
326 this.coll_names_list = valid_coll_names.toArray(coll_names_list);
327 return true;
328 }
329
330 protected Element processDocumentMetadataRetrieve(Element request)
331 {
332 // Create a new (empty) result message
333 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
334 result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
335 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
336
337 UserContext userContext = new UserContext(request);
338 // Get the parameters of the request
339 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
340 if (param_list == null)
341 {
342 logger.error("DocumentMetadataRetrieve request had no paramList.");
343 return result; // Return the empty result
344 }
345
346 NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
347 if (query_doc_list.getLength() == 0)
348 {
349 logger.error("DocumentMetadataRetrieve request had no documentNodes.");
350 return result; // Return the empty result
351 }
352
353 // the resulting doc node list
354 Element result_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
355 result.appendChild(result_node_list);
356
357 // get all the metadata params
358 Element new_param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
359 Element param = GSXML.createParameter(this.doc, "metadata", "Title");
360 new_param_list.appendChild(param);
361
362 // organise the nodes into collection lists
363 HashMap<String, Node> coll_map = new HashMap<String, Node>();
364
365 for (int i = 0; i < query_doc_list.getLength(); i++)
366 {
367 Element doc_node = (Element) query_doc_list.item(i);
368 String coll_name = doc_node.getAttribute("collection");
369 Element coll_items = (Element) coll_map.get(coll_name);
370 if (coll_items == null)
371 {
372 coll_items = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
373 coll_map.put(coll_name, coll_items);
374 }
375 coll_items.appendChild(this.doc.importNode(doc_node, true));
376 }
377
378 // create teh individual requests
379 Element meta_request_message = this.doc.createElement(GSXML.MESSAGE_ELEM);
380 Set mapping_set = coll_map.entrySet();
381 Iterator iter = mapping_set.iterator();
382
383 while (iter.hasNext())
384 {
385 Map.Entry e = (Map.Entry) iter.next();
386 String cname = (String) e.getKey();
387 Element doc_nodes = (Element) e.getValue();
388 Element meta_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), userContext);
389 meta_request.appendChild(doc_nodes);
390 meta_request.appendChild(new_param_list.cloneNode(true));
391 meta_request_message.appendChild(meta_request);
392
393 }
394
395 Node meta_result_node = this.router.process(meta_request_message);
396 Element meta_result = this.converter.nodeToElement(meta_result_node);
397
398 // now need to put the doc nodes back in the right order
399 // go through the original list again. keep an element pointer to
400 // the next element in each collections list
401 NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
402 for (int i = 0; i < meta_responses.getLength(); i++)
403 {
404 String collname = GSPath.removeLastLink(((Element) meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
405 Element first_elem = (Element) GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
406 coll_map.put(collname, first_elem);
407 }
408
409 for (int i = 0; i < query_doc_list.getLength(); i++)
410 {
411 Element doc_node = (Element) query_doc_list.item(i);
412 Element new_node = (Element) this.doc.importNode(doc_node, false);
413 result_node_list.appendChild(new_node);
414 String coll_name = doc_node.getAttribute("collection");
415
416 Element meta_elem = (Element) coll_map.get(coll_name);
417 GSXML.mergeMetadataLists(new_node, meta_elem);
418 coll_map.put(coll_name, meta_elem.getNextSibling());
419 }
420 return result;
421 }
422}
Note: See TracBrowser for help on using the repository browser.