1 | /*
|
---|
2 | * GS2LuceneSearch.java
|
---|
3 | * Copyright (C) 2006 New Zealand Digital Library, http://www.nzdl.org
|
---|
4 | *
|
---|
5 | * This program is free software; you can redistribute it and/or modify
|
---|
6 | * the Free Software Foundation; either version 2 of the License, or
|
---|
7 | * (at your option) any later version.
|
---|
8 | *
|
---|
9 | * This program is distributed in the hope that it will be useful,
|
---|
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
12 | * GNU General Public License for more details.
|
---|
13 | *
|
---|
14 | * You should have received a copy of the GNU General Public License
|
---|
15 | * along with this program; if not, write to the Free Software
|
---|
16 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
17 | */
|
---|
18 |
|
---|
19 | package org.greenstone.gsdl3.service;
|
---|
20 |
|
---|
21 | // Greenstone classes
|
---|
22 | import java.io.File;
|
---|
23 | import java.io.IOException;
|
---|
24 | import java.io.Serializable;
|
---|
25 | import java.util.ArrayList;
|
---|
26 | import java.util.HashMap;
|
---|
27 | import java.util.Iterator;
|
---|
28 | import java.util.List;
|
---|
29 | import java.util.Map;
|
---|
30 | import java.util.Set;
|
---|
31 | import java.util.Vector;
|
---|
32 |
|
---|
33 | // For maintaining Lucene IndexReader objects at collection level
|
---|
34 | import org.apache.lucene.index.DirectoryReader;
|
---|
35 | import org.apache.lucene.index.IndexReader;
|
---|
36 | import org.apache.lucene.store.Directory;
|
---|
37 | import org.apache.lucene.store.FSDirectory;
|
---|
38 |
|
---|
39 | import org.apache.log4j.Logger;
|
---|
40 | import org.greenstone.LuceneWrapper4.GS2LuceneQuery;
|
---|
41 | import org.greenstone.LuceneWrapper4.LuceneQueryResult;
|
---|
42 | import org.greenstone.gsdl3.util.FacetWrapper;
|
---|
43 | import org.greenstone.gsdl3.util.GSFile;
|
---|
44 | import org.greenstone.gsdl3.util.GSXML;
|
---|
45 | import org.greenstone.gsdl3.util.XMLConverter;
|
---|
46 | import org.w3c.dom.Document;
|
---|
47 | import org.w3c.dom.Element;
|
---|
48 |
|
---|
49 |
|
---|
50 | public class GS2LuceneSearch extends SharedSoleneGS2FieldSearch
|
---|
51 | {
|
---|
52 |
|
---|
53 | protected static final String SORT_ORDER_PARAM = "reverseSort";
|
---|
54 | protected static final String SORT_ORDER_REVERSE = "1";
|
---|
55 | protected static final String SORT_ORDER_NORMAL = "0";
|
---|
56 |
|
---|
57 | // IndexReader objects are to be opened for each index level (e.g. one for didx, one for sidx) of a
|
---|
58 | // collection and will live for the duration of that collection, which is from collection activation
|
---|
59 | // until deactivation.
|
---|
60 | // So we want singletons of each index level's IndexReader, since IndexReaders are "multi-threaded
|
---|
61 | // re-entrant", so there's support for just one reader per index with concurrent access by multiple users'
|
---|
62 | // search queries.
|
---|
63 | // When a collection is deactivated, we need to close the reader objects to prevent handles to the
|
---|
64 | // index lingering and causing file locking issues on windows.
|
---|
65 | // Since GS2LuceneQuery now becomes a local member variable instantiated per query, we have to maintain
|
---|
66 | // IndexReader objects in GS2LuceneSearch instead, as GS2LuceneSearch is a collection's service, and
|
---|
67 | // therefore activated and deactivated along with the collection.
|
---|
68 | // The uniqueness of an IndexReader is indicated in the filepath to its index folder (collection path + sidx/didx).
|
---|
69 | // It doesn't have to be a static map of index_dir to IndexReader, and can be a member variable, since
|
---|
70 | // no other collection will refer to the same didx and sidx index folders: each collection has unique filepaths
|
---|
71 | // to its collection folder's index subdirs, not shared with other collections so the Readers don't have to be
|
---|
72 | // shared between collections either.
|
---|
73 |
|
---|
74 | // We now store IndexReaders in a map of singleton index_dir -> IndexReaders opened for this collection:
|
---|
75 | // one Reader singleton for each index_dir
|
---|
76 | private Map<String, IndexReader> index_to_reader_map = new HashMap<String, IndexReader>();
|
---|
77 |
|
---|
78 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.GS2LuceneSearch.class.getName());
|
---|
79 |
|
---|
80 | public GS2LuceneSearch()
|
---|
81 | {
|
---|
82 | does_paging = true;
|
---|
83 | paramDefaults.put(SORT_ORDER_PARAM, SORT_ORDER_NORMAL);
|
---|
84 | }
|
---|
85 |
|
---|
86 | public void cleanUp()
|
---|
87 | {
|
---|
88 | super.cleanUp();
|
---|
89 |
|
---|
90 | // Prevent file locking issues: close all IndexReader objects maintained for this collection
|
---|
91 | synchronized(index_to_reader_map) { // Regular Map implementations are not synchronized, so adding/removing requires synchronizing on the map object.
|
---|
92 | // see https://docs.oracle.com/javase/7/docs/api/java/util/HashMap.html
|
---|
93 | // And ConcurrentHashMap seems complicated, https://docs.oracle.com/javase/7/docs/api/java/util/concurrent/ConcurrentHashMap.html
|
---|
94 |
|
---|
95 | // Synchronizing *outside* the loop because cleanUp() clears the entire HashMap.
|
---|
96 | // Don't let any other threads access the map, hence synchronizing.
|
---|
97 | // Not sure if there may be other threads accessing the map when deactivating a collection which calls cleanUp().
|
---|
98 | // However, when multiple users' search queries lead to adding to the hashmap, definitely need to
|
---|
99 | // synchronize as there's a greater possibility of concurrent access then.
|
---|
100 |
|
---|
101 | Iterator<Map.Entry<String,IndexReader>> map_iterator = index_to_reader_map.entrySet().iterator();
|
---|
102 | // Can use the Map.Entry Set view iterator to remove (key, value) entry from underlying Map!
|
---|
103 | // See https://docs.oracle.com/javase/7/docs/api/java/util/HashMap.html#keySet()
|
---|
104 | // Same thread creates the iterator as synchronizes on the map, so we should be allowed to remove() from the map
|
---|
105 | // but only through iterator!
|
---|
106 | while(map_iterator.hasNext()) {
|
---|
107 | Map.Entry<String,IndexReader> entry = map_iterator.next();
|
---|
108 | //index_to_reader_map.remove(...); // concurrentmodexception! Only allowed to remove through iterator. Will remove recent object returned by next()
|
---|
109 | IndexReader reader = entry.getValue(); //keys are index dir paths, e.g. path to current collection's didx folder, values are IndexReader objects
|
---|
110 | map_iterator.remove(); // removes current key's (key,value) entry from underlying map! (Remember, we're iterating on the keyset)
|
---|
111 | // We're first removing the reader singleton from map because reader.close() will only close the reader
|
---|
112 | //if it's the final reference to it in case that has a bearing here
|
---|
113 |
|
---|
114 | if(reader != null) { // if there was a reader singleton instantiated for this index directory, e.g. coll-didx, close it
|
---|
115 | try {
|
---|
116 | // We're opening an IndexReader per indexdir once and closing it once: at start and end of collection.
|
---|
117 | // If Reader was a member var of GS2LuceneQuery and if multiple GS2LuceneQuery Objects were to call close() on the
|
---|
118 | // same reader object (on the singleton instance of reader for an index dir), so close is called multiple times,
|
---|
119 | // then would use incRef and decRef, see http://lucene.472066.n3.nabble.com/IndexReader-close-behavior-td2865515.html
|
---|
120 | // But then when concurrent queries are done, the final one would have closed the IndexReader and it would have to
|
---|
121 | // be reopened for the next query. We'd rather keep an opened IndexReader around until the collection's deactivated.
|
---|
122 | reader.close();
|
---|
123 | // Closes files associated with this index. Also saves any new deletions to disk.
|
---|
124 | // No other methods should be called after this has been called.
|
---|
125 | } catch (IOException exception) {
|
---|
126 | exception.printStackTrace();
|
---|
127 | }
|
---|
128 | }
|
---|
129 | } // end loop
|
---|
130 | } // end synchronising on index_to_reader_map
|
---|
131 |
|
---|
132 | // Now we've closed all the Readers maintained for this collection and cleared the map.
|
---|
133 | }
|
---|
134 |
|
---|
135 | public boolean configure(Element info, Element extra_info)
|
---|
136 | {
|
---|
137 | if (!super.configure(info, extra_info))
|
---|
138 | {
|
---|
139 | return false;
|
---|
140 | }
|
---|
141 | logger.info("Configuring GS2LuceneSearch...");
|
---|
142 |
|
---|
143 | // add our reverseSort param to be saved to the session
|
---|
144 | this.save_params.add(SORT_ORDER_PARAM);
|
---|
145 | return true;
|
---|
146 | }
|
---|
147 | /** add in the Lucene specific params to TextQuery */
|
---|
148 | protected void addCustomQueryParams(Element param_list, String lang)
|
---|
149 | {
|
---|
150 | super.addCustomQueryParams(param_list, lang);
|
---|
151 | /** Add in the reverse sort on/off param */
|
---|
152 | createParameter(SORT_ORDER_PARAM, param_list, lang);
|
---|
153 | }
|
---|
154 | /** add in Lucene specific params for AdvancedFieldQuery */
|
---|
155 | protected void addCustomQueryParamsAdvField(Element param_list, String lang)
|
---|
156 | {
|
---|
157 | super.addCustomQueryParamsAdvField(param_list, lang);
|
---|
158 | createParameter(SORT_ORDER_PARAM, param_list, lang);
|
---|
159 |
|
---|
160 | }
|
---|
161 | /** create a param and add to the list */
|
---|
162 | protected void createParameter(String name, Element param_list, String lang)
|
---|
163 | {
|
---|
164 | Document doc = param_list.getOwnerDocument();
|
---|
165 | Element param = null;
|
---|
166 | String param_default = paramDefaults.get(name);
|
---|
167 | if (name.equals(SORT_ORDER_PARAM)) {
|
---|
168 | String[] vals = { SORT_ORDER_REVERSE, SORT_ORDER_NORMAL };
|
---|
169 | String[] vals_texts = { getTextString("param." + SORT_ORDER_PARAM + "." + SORT_ORDER_REVERSE, lang), getTextString("param." + SORT_ORDER_PARAM + "." + SORT_ORDER_NORMAL, lang) };
|
---|
170 |
|
---|
171 | param = GSXML.createParameterDescription(doc, SORT_ORDER_PARAM, getTextString("param." + SORT_ORDER_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, param_default, vals, vals_texts);
|
---|
172 | }
|
---|
173 |
|
---|
174 | if (param != null)
|
---|
175 | {
|
---|
176 | param_list.appendChild(param);
|
---|
177 | }
|
---|
178 | else
|
---|
179 | {
|
---|
180 | super.createParameter(name, param_list, lang);
|
---|
181 | }
|
---|
182 |
|
---|
183 | }
|
---|
184 |
|
---|
185 | /** methods to handle actually doing the query */
|
---|
186 |
|
---|
187 | /** do any initialisation of the query object */
|
---|
188 | protected Object setUpQueryer(HashMap params)
|
---|
189 | {
|
---|
190 | // local Query object
|
---|
191 | GS2LuceneQuery lucene_src = new GS2LuceneQuery();
|
---|
192 |
|
---|
193 | String indexdir = GSFile.collectionBaseDir(this.site_home, this.cluster_name) + File.separatorChar + "index" + File.separatorChar;
|
---|
194 |
|
---|
195 | String index = "didx";
|
---|
196 | if (this.default_level.toUpperCase().equals("SEC")) {
|
---|
197 | index = "sidx";
|
---|
198 | }
|
---|
199 | String physical_index_language_name = null;
|
---|
200 | String physical_sub_index_name = null;
|
---|
201 | int hits_per_page = Integer.parseInt(paramDefaults.get(HITS_PER_PAGE_PARAM));
|
---|
202 | int start_page = Integer.parseInt(paramDefaults.get(START_PAGE_PARAM));
|
---|
203 | String sort_field = getLuceneSort(default_sort);
|
---|
204 | String sort_order = paramDefaults.get(SORT_ORDER_PARAM);
|
---|
205 |
|
---|
206 | // set up the query params
|
---|
207 | Set entries = params.entrySet();
|
---|
208 | Iterator i = entries.iterator();
|
---|
209 | while (i.hasNext())
|
---|
210 | {
|
---|
211 | Map.Entry m = (Map.Entry) i.next();
|
---|
212 | String name = (String) m.getKey();
|
---|
213 | String value = (String) m.getValue();
|
---|
214 |
|
---|
215 | if (name.equals(HITS_PER_PAGE_PARAM))
|
---|
216 | {
|
---|
217 | if (value.equals("all")) {
|
---|
218 | hits_per_page = -1;
|
---|
219 | } else {
|
---|
220 | hits_per_page = Integer.parseInt(value);
|
---|
221 | }
|
---|
222 | }
|
---|
223 | else if (name.equals(START_PAGE_PARAM))
|
---|
224 | {
|
---|
225 | start_page = Integer.parseInt(value);
|
---|
226 |
|
---|
227 | }
|
---|
228 | else if (name.equals(MATCH_PARAM))
|
---|
229 | {
|
---|
230 | if (value.equals(MATCH_PARAM_ALL))
|
---|
231 | {
|
---|
232 | lucene_src.setDefaultConjunctionOperator("AND");
|
---|
233 | }
|
---|
234 | else
|
---|
235 | {
|
---|
236 | lucene_src.setDefaultConjunctionOperator("OR");
|
---|
237 | }
|
---|
238 | }
|
---|
239 | else if (name.equals(RANK_PARAM))
|
---|
240 | {
|
---|
241 | sort_field = getLuceneSort(value);
|
---|
242 | lucene_src.setSortField(sort_field);
|
---|
243 |
|
---|
244 | }
|
---|
245 | else if (name.equals(SORT_ORDER_PARAM)) {
|
---|
246 | sort_order = value;
|
---|
247 | }
|
---|
248 | else if (name.equals(LEVEL_PARAM))
|
---|
249 | {
|
---|
250 | if (value.toUpperCase().equals("SEC"))
|
---|
251 | {
|
---|
252 | index = "sidx";
|
---|
253 | }
|
---|
254 | else
|
---|
255 | {
|
---|
256 | index = "didx";
|
---|
257 | }
|
---|
258 | }
|
---|
259 | else if (name.equals(INDEX_SUBCOLLECTION_PARAM))
|
---|
260 | {
|
---|
261 | physical_sub_index_name = value;
|
---|
262 | }
|
---|
263 | else if (name.equals(INDEX_LANGUAGE_PARAM))
|
---|
264 | {
|
---|
265 | physical_index_language_name = value;
|
---|
266 | } // ignore any others
|
---|
267 | }
|
---|
268 | // set up start and end results if necessary
|
---|
269 | // start results always start at 0
|
---|
270 | int start_results = 0;
|
---|
271 | if (start_page > 1 && hits_per_page > 0)
|
---|
272 | {
|
---|
273 | start_results = ((start_page - 1) * hits_per_page) ;
|
---|
274 | }
|
---|
275 | int end_results = Integer.MAX_VALUE;
|
---|
276 | if (hits_per_page > 0) {
|
---|
277 | end_results = hits_per_page * start_page;
|
---|
278 | }
|
---|
279 | lucene_src.setStartResults(start_results);
|
---|
280 | lucene_src.setEndResults(end_results);
|
---|
281 |
|
---|
282 | if (index.equals("sidx") || index.equals("didx"))
|
---|
283 | {
|
---|
284 | if (physical_sub_index_name != null)
|
---|
285 | {
|
---|
286 | index += physical_sub_index_name;
|
---|
287 | }
|
---|
288 | if (physical_index_language_name != null)
|
---|
289 | {
|
---|
290 | index += physical_index_language_name;
|
---|
291 | }
|
---|
292 | }
|
---|
293 |
|
---|
294 | if (sort_order.equals(SORT_ORDER_REVERSE)) {
|
---|
295 | lucene_src.setReverseSort(true);
|
---|
296 | } else {
|
---|
297 | lucene_src.setReverseSort(false);
|
---|
298 | }
|
---|
299 |
|
---|
300 | String full_index_dir_str = indexdir + index;
|
---|
301 | lucene_src.setIndexDir(full_index_dir_str);
|
---|
302 |
|
---|
303 | // Ensure we have an IndexReader for this full_index_dir_str:
|
---|
304 | // check the hashmap first, in case we already opened a reader and searcher for this index dir, e.g. didx
|
---|
305 | // if there was a reader singleton instantiated for this index directory, e.g. <coll>didx, use that.
|
---|
306 | // Else open a new reader for this index_dir and store it in the map.
|
---|
307 | IndexReader reader = index_to_reader_map.get(full_index_dir_str);
|
---|
308 | if(reader == null) {
|
---|
309 | try {
|
---|
310 | Directory full_indexdir_dir = FSDirectory.open(new File(full_index_dir_str));
|
---|
311 | reader = DirectoryReader.open(full_indexdir_dir); // Returns an IndexReader reading the index in the given Directory. now readOnly=true by default, and therefore also for searcher
|
---|
312 | synchronized(index_to_reader_map) {
|
---|
313 | // If storing searcher along with reader, mimic Pairs with: https://stackoverflow.com/questions/2670982/using-pairs-or-2-tuples-in-java
|
---|
314 | index_to_reader_map.put(full_index_dir_str, reader);
|
---|
315 | }
|
---|
316 | }
|
---|
317 | catch (IOException exception) {
|
---|
318 | exception.printStackTrace();
|
---|
319 | }
|
---|
320 | }
|
---|
321 |
|
---|
322 | lucene_src.initialise(reader); // sets IndexReader and IndexSearcher
|
---|
323 |
|
---|
324 | return lucene_src; // return the queryobject
|
---|
325 | }
|
---|
326 |
|
---|
327 | /** do the query */
|
---|
328 | protected Object runQuery(Object queryObject, String query)
|
---|
329 | {
|
---|
330 | GS2LuceneQuery lucene_src = (GS2LuceneQuery) queryObject;
|
---|
331 | try
|
---|
332 | {
|
---|
333 | LuceneQueryResult lqr = lucene_src.runQuery(query);
|
---|
334 | return lqr;
|
---|
335 | }
|
---|
336 | catch (Exception e)
|
---|
337 | {
|
---|
338 | logger.error("Exception happened in runQuery(): ", e);
|
---|
339 | }
|
---|
340 |
|
---|
341 | return null;
|
---|
342 | }
|
---|
343 |
|
---|
344 | /** get the total number of docs that match */
|
---|
345 | protected long numDocsMatched(Object query_result)
|
---|
346 | {
|
---|
347 | return ((LuceneQueryResult) query_result).getTotalDocs();
|
---|
348 | }
|
---|
349 |
|
---|
350 | /** get the list of doc ids */
|
---|
351 | protected String[] getDocIDs(Object query_result)
|
---|
352 | {
|
---|
353 | Vector docs = ((LuceneQueryResult) query_result).getDocs();
|
---|
354 | String[] doc_nums = new String[docs.size()];
|
---|
355 | for (int d = 0; d < docs.size(); d++)
|
---|
356 | {
|
---|
357 | String doc_num = ((LuceneQueryResult.DocInfo) docs.elementAt(d)).id_;
|
---|
358 | doc_nums[d] = doc_num;
|
---|
359 | }
|
---|
360 | return doc_nums;
|
---|
361 | }
|
---|
362 |
|
---|
363 | /** get the list of doc ranks */
|
---|
364 | protected String[] getDocRanks(Object query_result)
|
---|
365 | {
|
---|
366 | Vector docs = ((LuceneQueryResult) query_result).getDocs();
|
---|
367 | String[] doc_ranks = new String[docs.size()];
|
---|
368 | for (int d = 0; d < docs.size(); d++)
|
---|
369 | {
|
---|
370 | doc_ranks[d] = Float.toString(((LuceneQueryResult.DocInfo) docs.elementAt(d)).rank_);
|
---|
371 | }
|
---|
372 | return doc_ranks;
|
---|
373 | }
|
---|
374 |
|
---|
375 | /** add in term info if available */
|
---|
376 | protected boolean addTermInfo(Element term_list, HashMap params, Object query_result)
|
---|
377 | {
|
---|
378 | Document doc = term_list.getOwnerDocument();
|
---|
379 | String query_level = (String) params.get(LEVEL_PARAM); // the current query level
|
---|
380 |
|
---|
381 | Vector terms = ((LuceneQueryResult) query_result).getTerms();
|
---|
382 | for (int t = 0; t < terms.size(); t++)
|
---|
383 | {
|
---|
384 | LuceneQueryResult.TermInfo term_info = (LuceneQueryResult.TermInfo) terms.get(t);
|
---|
385 |
|
---|
386 | Element term_elem = doc.createElement(GSXML.TERM_ELEM);
|
---|
387 | term_elem.setAttribute(GSXML.NAME_ATT, term_info.term_);
|
---|
388 | term_elem.setAttribute(FREQ_ATT, "" + term_info.term_freq_);
|
---|
389 | term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "" + term_info.match_docs_);
|
---|
390 | term_elem.setAttribute(FIELD_ATT, term_info.field_);
|
---|
391 | term_list.appendChild(term_elem);
|
---|
392 | }
|
---|
393 |
|
---|
394 | Vector stopwords = ((LuceneQueryResult) query_result).getStopWords();
|
---|
395 | for (int t = 0; t < stopwords.size(); t++)
|
---|
396 | {
|
---|
397 | String stopword = (String) stopwords.get(t);
|
---|
398 |
|
---|
399 | Element stopword_elem = doc.createElement(GSXML.STOPWORD_ELEM);
|
---|
400 | stopword_elem.setAttribute(GSXML.NAME_ATT, stopword);
|
---|
401 | term_list.appendChild(stopword_elem);
|
---|
402 | }
|
---|
403 |
|
---|
404 | return true;
|
---|
405 | }
|
---|
406 |
|
---|
407 | protected ArrayList<FacetWrapper> getFacets(Object query_result, String lang)
|
---|
408 | {
|
---|
409 | return null;
|
---|
410 | }
|
---|
411 |
|
---|
412 | protected String getLuceneSort(String gs3_sort) {
|
---|
413 |
|
---|
414 | if (gs3_sort.equals(RANK_PARAM_RANK)) {
|
---|
415 | return GS2LuceneQuery.SORT_RANK;
|
---|
416 | }
|
---|
417 | if (gs3_sort.equals(RANK_PARAM_NONE)) {
|
---|
418 | return GS2LuceneQuery.SORT_NATURAL;
|
---|
419 | }
|
---|
420 | return gs3_sort;
|
---|
421 | }
|
---|
422 |
|
---|
423 | @Override
|
---|
424 | protected Map<String, Map<String, List<String>>> getHighlightSnippets(
|
---|
425 | Object query_result) {
|
---|
426 | // TODO Auto-generated method stub
|
---|
427 | return null;
|
---|
428 | }
|
---|
429 |
|
---|
430 | }
|
---|