1 | /**********************************************************************
|
---|
2 | *
|
---|
3 | * GS2LuceneQuery.java
|
---|
4 | *
|
---|
5 | * Copyright 2004 The New Zealand Digital Library Project
|
---|
6 | *
|
---|
7 | * A component of the Greenstone digital library software
|
---|
8 | * from the New Zealand Digital Library Project at the
|
---|
9 | * University of Waikato, New Zealand.
|
---|
10 | *
|
---|
11 | * This program is free software; you can redistribute it and/or modify
|
---|
12 | * it under the terms of the GNU General Public License as published by
|
---|
13 | * the Free Software Foundation; either version 2 of the License, or
|
---|
14 | * (at your option) any later version.
|
---|
15 | *
|
---|
16 | * This program is distributed in the hope that it will be useful,
|
---|
17 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
18 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
19 | * GNU General Public License for more details.
|
---|
20 | *
|
---|
21 | * You should have received a copy of the GNU General Public License
|
---|
22 | * along with this program; if not, write to the Free Software
|
---|
23 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
24 | *
|
---|
25 | *********************************************************************/
|
---|
26 | package org.greenstone.LuceneWrapper4;
|
---|
27 |
|
---|
28 |
|
---|
29 | import java.io.*;
|
---|
30 | import java.util.*;
|
---|
31 | import java.util.regex.*;
|
---|
32 |
|
---|
33 | import org.apache.lucene.analysis.Analyzer;
|
---|
34 | import org.apache.lucene.analysis.standard.StandardAnalyzer;
|
---|
35 | import org.apache.lucene.document.Document;
|
---|
36 | import org.apache.lucene.index.DirectoryReader;
|
---|
37 | import org.apache.lucene.index.IndexReader;
|
---|
38 | import org.apache.lucene.index.Term;
|
---|
39 | //import org.apache.lucene.index.TermDocs;
|
---|
40 | import org.apache.lucene.queryparser.classic.ParseException;
|
---|
41 | import org.apache.lucene.queryparser.classic.QueryParser;
|
---|
42 | import org.apache.lucene.search.BooleanClause;
|
---|
43 | import org.apache.lucene.search.BooleanQuery; // for the TooManyClauses exception
|
---|
44 | import org.apache.lucene.search.ConstantScoreQuery;
|
---|
45 | import org.apache.lucene.search.Filter;
|
---|
46 | import org.apache.lucene.search.IndexSearcher;
|
---|
47 | import org.apache.lucene.search.MultiTermQuery;
|
---|
48 | import org.apache.lucene.search.MultiTermQuery.ConstantScoreAutoRewrite;
|
---|
49 | import org.apache.lucene.search.Query;
|
---|
50 | import org.apache.lucene.search.TermRangeFilter;
|
---|
51 | import org.apache.lucene.search.IndexSearcher; // Searcher is deprecated
|
---|
52 | import org.apache.lucene.search.ScoreDoc;
|
---|
53 | import org.apache.lucene.search.Sort;
|
---|
54 | import org.apache.lucene.search.SortField;
|
---|
55 | import org.apache.lucene.search.TopFieldDocs;
|
---|
56 |
|
---|
57 | import org.apache.lucene.index.DocsEnum;
|
---|
58 | import org.apache.lucene.index.MultiFields;
|
---|
59 |
|
---|
60 | import org.apache.lucene.store.Directory;
|
---|
61 | import org.apache.lucene.store.FSDirectory;
|
---|
62 |
|
---|
63 | import org.apache.lucene.util.Bits;
|
---|
64 | import org.apache.lucene.util.BytesRef;
|
---|
65 | import org.apache.lucene.util.Version;
|
---|
66 |
|
---|
67 | public class GS2LuceneQuery extends SharedSoleneQuery
|
---|
68 | {
|
---|
69 | public static String SORT_RANK = "rank";
|
---|
70 | public static String SORT_NATURAL = "natural";
|
---|
71 |
|
---|
72 | protected String full_indexdir="";
|
---|
73 |
|
---|
74 | protected SortField.Type sort_type = SortField.Type.SCORE;
|
---|
75 | protected boolean reverse_sort = false;
|
---|
76 | protected Sort sorter=new Sort();
|
---|
77 | protected Filter filter = null;
|
---|
78 |
|
---|
79 | protected QueryParser query_parser = null;
|
---|
80 | protected QueryParser query_parser_no_stop_words = null;
|
---|
81 | protected IndexSearcher searcher = null;
|
---|
82 | protected IndexReader reader = null;
|
---|
83 |
|
---|
84 | public GS2LuceneQuery() {
|
---|
85 | super();
|
---|
86 |
|
---|
87 | // Create one query parser with the standard set of stop words, and one with none
|
---|
88 |
|
---|
89 | query_parser = new QueryParser(GSLuceneConstants.MATCH_VERSION, TEXTFIELD, new GS2Analyzer()); // uses built-in stop_words_set
|
---|
90 | query_parser_no_stop_words = new QueryParser(GSLuceneConstants.MATCH_VERSION, TEXTFIELD, new GS2Analyzer(new String[] { }));
|
---|
91 | }
|
---|
92 |
|
---|
93 |
|
---|
94 | public boolean initialise() {
|
---|
95 |
|
---|
96 | if (!super.initialise()) {
|
---|
97 | return false;
|
---|
98 | }
|
---|
99 |
|
---|
100 |
|
---|
101 | if (full_indexdir==null || full_indexdir.length()==-1){
|
---|
102 | utf8out.println("Index directory is not indicated ");
|
---|
103 | utf8out.flush();
|
---|
104 | return false;
|
---|
105 | }
|
---|
106 |
|
---|
107 | try {
|
---|
108 |
|
---|
109 | if(reader != null) {
|
---|
110 | reader.close();
|
---|
111 | searcher = null;
|
---|
112 | }
|
---|
113 |
|
---|
114 | Directory full_indexdir_dir = FSDirectory.open(new File(full_indexdir));
|
---|
115 |
|
---|
116 | reader = DirectoryReader.open(full_indexdir_dir); // Returns a IndexReader reading the index in the given Directory. now readOnly=true by default, and therefore also for searcher
|
---|
117 | searcher = new IndexSearcher(reader); // during searcher.search() will get it to compute ranks when sorting by fields
|
---|
118 |
|
---|
119 | this.sorter = new Sort(new SortField(this.sort_field, this.sort_type, this.reverse_sort));
|
---|
120 | }
|
---|
121 | catch (IOException exception) {
|
---|
122 | exception.printStackTrace();
|
---|
123 | return false;
|
---|
124 | }
|
---|
125 | return true;
|
---|
126 |
|
---|
127 | }
|
---|
128 |
|
---|
129 | public void setIndexDir(String full_indexdir) {
|
---|
130 | this.full_indexdir = full_indexdir;
|
---|
131 | }
|
---|
132 |
|
---|
133 | public void setSortField(String sort_field) {
|
---|
134 | if (sort_field.equals(SORT_RANK)) {
|
---|
135 | this.sort_field = null;
|
---|
136 | this.sort_type = SortField.Type.SCORE;
|
---|
137 | } else if (sort_field.equals(SORT_NATURAL)) {
|
---|
138 | this.sort_field = null;
|
---|
139 | this.sort_type = SortField.Type.DOC;
|
---|
140 | } else {
|
---|
141 | this.sort_field = sort_field;
|
---|
142 | this.sort_type = SortField.Type.STRING; // for now. numeric??
|
---|
143 | }
|
---|
144 | }
|
---|
145 | public void setReverseSort(boolean reverse) {
|
---|
146 | this.reverse_sort = reverse;
|
---|
147 | }
|
---|
148 | public boolean getReverseSort() {
|
---|
149 | return this.reverse_sort;
|
---|
150 | }
|
---|
151 |
|
---|
152 | public void setFilterString(String filter_string) {
|
---|
153 | super.setFilterString(filter_string);
|
---|
154 | this.filter = parseFilterString(filter_string);
|
---|
155 | }
|
---|
156 |
|
---|
157 | public Filter getFilter() {
|
---|
158 | return this.filter;
|
---|
159 | }
|
---|
160 |
|
---|
161 |
|
---|
162 | public LuceneQueryResult runQuery(String query_string) {
|
---|
163 |
|
---|
164 | if (query_string == null || query_string.equals("")) {
|
---|
165 | utf8out.println("The query word is not indicated ");
|
---|
166 | utf8out.flush();
|
---|
167 | return null;
|
---|
168 | }
|
---|
169 |
|
---|
170 | LuceneQueryResult lucene_query_result=new LuceneQueryResult();
|
---|
171 | lucene_query_result.clear();
|
---|
172 |
|
---|
173 | try {
|
---|
174 | Query query_including_stop_words = query_parser_no_stop_words.parse(query_string);
|
---|
175 | query_including_stop_words = query_including_stop_words.rewrite(reader);
|
---|
176 |
|
---|
177 | // System.err.println("********* query_string " + query_string + "****");
|
---|
178 |
|
---|
179 | Query query = parseQuery(reader, query_parser, query_string, fuzziness);
|
---|
180 | query = recursivelyRewriteQuery(query, reader, lucene_query_result);
|
---|
181 | // System.err.println("@@@@ final query class name: " + query.getClass());
|
---|
182 |
|
---|
183 | // http://stackoverflow.com/questions/13537126/term-frequency-in-lucene-4-0
|
---|
184 | // http://stackoverflow.com/questions/20575254/lucene-4-4-how-to-get-term-frequency-over-all-index
|
---|
185 | // http://stackoverflow.com/questions/8938960/how-to-get-document-ids-for-document-term-vector-in-lucene?rq=1
|
---|
186 | // https://github.com/hibernate/hibernate-search/blob/master/orm/src/test/java/org/hibernate/search/test/filter/BestDriversFilter.java
|
---|
187 | // http://lucene.apache.org/core/4_7_2/MIGRATE.html
|
---|
188 |
|
---|
189 | // Get the list of expanded query terms and their frequencies
|
---|
190 | // num docs matching, and total frequency
|
---|
191 | HashSet terms = new HashSet();
|
---|
192 | query.extractTerms(terms);
|
---|
193 |
|
---|
194 | HashMap doc_term_freq_map = new HashMap();
|
---|
195 |
|
---|
196 | Iterator iter = terms.iterator();
|
---|
197 |
|
---|
198 | Bits liveDocs = null;
|
---|
199 | if(reader.hasDeletions()) {
|
---|
200 | System.err.println("@@@ GS2LuceneQuery.java: There have been deletions. Merging to get liveDocs.");
|
---|
201 | liveDocs = MultiFields.getLiveDocs(reader); // SLOW! But getLiveDocs returns null if there are no deletions
|
---|
202 | }
|
---|
203 |
|
---|
204 | while (iter.hasNext()) {
|
---|
205 |
|
---|
206 | // http://stackoverflow.com/questions/13537126/term-frequency-in-lucene-4-0
|
---|
207 |
|
---|
208 | Term term = (Term) iter.next();
|
---|
209 | // System.err.println("@@@ GS2LuceneQuery.java: Next term: " + term.text());
|
---|
210 | BytesRef term_bytes = term.bytes();
|
---|
211 | DocsEnum term_docs = MultiFields.getTermDocsEnum(reader, liveDocs, term.field(), term_bytes); // flags?
|
---|
212 |
|
---|
213 | // Get the term frequency over all the documents
|
---|
214 | //TermDocs term_docs = reader.termDocs(term);
|
---|
215 | int term_freq = 0;
|
---|
216 | int match_docs = 0;
|
---|
217 |
|
---|
218 | if(term_docs != null) {
|
---|
219 | int docID = -1;
|
---|
220 | while((docID = term_docs.nextDoc()) != DocsEnum.NO_MORE_DOCS) {//while (term_docs.next())
|
---|
221 | if (term_docs.freq() != 0)
|
---|
222 | {
|
---|
223 | term_freq += term_docs.freq();
|
---|
224 | match_docs++;
|
---|
225 |
|
---|
226 | // Calculate the document-level term frequency as well
|
---|
227 | Integer lucene_doc_num_obj = new Integer(term_docs.docID());
|
---|
228 | int doc_term_freq = 0;
|
---|
229 | if (doc_term_freq_map.containsKey(lucene_doc_num_obj))
|
---|
230 | {
|
---|
231 | doc_term_freq = ((Integer) doc_term_freq_map.get(lucene_doc_num_obj)).intValue();
|
---|
232 | }
|
---|
233 | doc_term_freq += term_docs.freq();
|
---|
234 |
|
---|
235 | doc_term_freq_map.put(lucene_doc_num_obj, new Integer(doc_term_freq));
|
---|
236 | }
|
---|
237 | }
|
---|
238 | } else {
|
---|
239 | System.err.println("@@@ GS2LuceneQuery.java: term_docs is null for term " + term.text());
|
---|
240 | }
|
---|
241 |
|
---|
242 | // Create a term
|
---|
243 | lucene_query_result.addTerm(term.text(), term.field(), match_docs, term_freq);
|
---|
244 | }
|
---|
245 |
|
---|
246 | // Get the list of stop words removed from the query
|
---|
247 | HashSet terms_including_stop_words = new HashSet();
|
---|
248 | query_including_stop_words.extractTerms(terms_including_stop_words);
|
---|
249 | Iterator terms_including_stop_words_iter = terms_including_stop_words.iterator();
|
---|
250 | while (terms_including_stop_words_iter.hasNext()) {
|
---|
251 | Term term = (Term) terms_including_stop_words_iter.next();
|
---|
252 | if (!terms.contains(term)) {
|
---|
253 | lucene_query_result.addStopWord(term.text());
|
---|
254 | }
|
---|
255 | }
|
---|
256 |
|
---|
257 | // Extracting all documents for a given search - http://www.gossamer-threads.com/lists/lucene/java-user/134873
|
---|
258 | // http://lucene.apache.org/core/3_4_0/api/core/org/apache/lucene/search/TotalHitCountCollector.html
|
---|
259 | // http://lucene.apache.org/core/4_7_2/core/index.html?org/apache/lucene/search/TopFieldDocs.html
|
---|
260 |
|
---|
261 | // 1. Figure out how many results there will be.
|
---|
262 | //TotalHitCountCollecter countCollector = new TotalHitCountCollector();
|
---|
263 | //searcher.search(query, filter, collector);
|
---|
264 | //int hitCount = collector.count;
|
---|
265 |
|
---|
266 | // Actually do the query
|
---|
267 | // Simple case for getting all the matching documents
|
---|
268 | if (end_results == Integer.MAX_VALUE) {
|
---|
269 | // Perform the query (filter and sorter may be null)
|
---|
270 | TopFieldDocs hits = searcher.search(query, filter, end_results, sorter, true, true); // doDocScores=true, doMaxScore=true
|
---|
271 | // Is there a slight difference in the definition between
|
---|
272 | // https://lucene.apache.org/core/3_6_0/api/all/org/apache/lucene/search/IndexSearcher.html#setDefaultFieldSortScoring%28boolean,%20boolean%29
|
---|
273 | // and http://lucene.apache.org/core/4_7_2/core/org/apache/lucene/search/IndexSearcher.html#search%28org.apache.lucene.search.Query,%20org.apache.lucene.search.Filter,%20int,%20org.apache.lucene.search.Sort,%20boolean,%20boolean%29
|
---|
274 | // Seems to be okay.
|
---|
275 | // See also http://stackoverflow.com/questions/7910241/in-lucene-what-is-the-purpose-of-setdefaultfieldsortscoring
|
---|
276 |
|
---|
277 | lucene_query_result.setTotalDocs(hits.totalHits);
|
---|
278 |
|
---|
279 | // Output the matching documents
|
---|
280 | lucene_query_result.setStartResults(start_results);
|
---|
281 | lucene_query_result.setEndResults(hits.totalHits); // ??
|
---|
282 |
|
---|
283 | for (int i = start_results; i < hits.totalHits; i++) {
|
---|
284 | int lucene_doc_num = hits.scoreDocs[i ].doc; // i-1
|
---|
285 | Document doc = reader.document(lucene_doc_num);
|
---|
286 | int doc_term_freq = 0;
|
---|
287 | Integer doc_term_freq_object = (Integer) doc_term_freq_map.get(new Integer(lucene_doc_num));
|
---|
288 | if (doc_term_freq_object != null)
|
---|
289 | {
|
---|
290 | doc_term_freq = doc_term_freq_object.intValue();
|
---|
291 | }
|
---|
292 | lucene_query_result.addDoc(doc.get("docOID").trim(), hits.scoreDocs[i].score, doc_term_freq);
|
---|
293 | }
|
---|
294 | }
|
---|
295 |
|
---|
296 | // Slightly more complicated case for returning a subset of the matching documents
|
---|
297 | else {
|
---|
298 | // Perform the query (filter may be null)
|
---|
299 | TopFieldDocs hits = searcher.search(query, filter, end_results, sorter, true, true); // doDocScores=true, doMaxScore=true
|
---|
300 | // See also http://stackoverflow.com/questions/7910241/in-lucene-what-is-the-purpose-of-setdefaultfieldsortscoring
|
---|
301 | lucene_query_result.setTotalDocs(hits.totalHits);
|
---|
302 |
|
---|
303 | lucene_query_result.setStartResults(start_results);
|
---|
304 | lucene_query_result.setEndResults(end_results < hits.scoreDocs.length ? end_results: hits.scoreDocs.length);
|
---|
305 |
|
---|
306 | // Output the matching documents
|
---|
307 | for (int i = start_results; (i < hits.scoreDocs.length && i < end_results); i++) {
|
---|
308 | int lucene_doc_num = hits.scoreDocs[i].doc;
|
---|
309 | Document doc = reader.document(lucene_doc_num);
|
---|
310 | int doc_term_freq = 0;
|
---|
311 | Integer doc_term_freq_object = (Integer) doc_term_freq_map.get(new Integer(lucene_doc_num));
|
---|
312 | if (doc_term_freq_object != null)
|
---|
313 | {
|
---|
314 | doc_term_freq = doc_term_freq_object.intValue();
|
---|
315 | }
|
---|
316 | lucene_query_result.addDoc(doc.get("docOID").trim(), hits.scoreDocs[i].score, doc_term_freq);
|
---|
317 | }
|
---|
318 | }
|
---|
319 | }
|
---|
320 |
|
---|
321 | catch (ParseException parse_exception) {
|
---|
322 | lucene_query_result.setError(LuceneQueryResult.PARSE_ERROR);
|
---|
323 | }
|
---|
324 | catch (BooleanQuery.TooManyClauses too_many_clauses_exception) {
|
---|
325 | lucene_query_result.setError(LuceneQueryResult.TOO_MANY_CLAUSES_ERROR);
|
---|
326 | }
|
---|
327 | catch (IOException exception) {
|
---|
328 | lucene_query_result.setError(LuceneQueryResult.IO_ERROR);
|
---|
329 | exception.printStackTrace();
|
---|
330 | }
|
---|
331 | catch (Exception exception) {
|
---|
332 | lucene_query_result.setError(LuceneQueryResult.OTHER_ERROR);
|
---|
333 | exception.printStackTrace();
|
---|
334 | }
|
---|
335 | return lucene_query_result;
|
---|
336 | }
|
---|
337 |
|
---|
338 | public void setDefaultConjunctionOperator(String default_conjunction_operator) {
|
---|
339 | super.setDefaultConjunctionOperator(default_conjunction_operator);
|
---|
340 |
|
---|
341 | if (default_conjunction_operator.equals("AND")) {
|
---|
342 | query_parser.setDefaultOperator(query_parser.AND_OPERATOR);
|
---|
343 | query_parser_no_stop_words.setDefaultOperator(query_parser.AND_OPERATOR);
|
---|
344 | } else { // default is OR
|
---|
345 | query_parser.setDefaultOperator(query_parser.OR_OPERATOR);
|
---|
346 | query_parser_no_stop_words.setDefaultOperator(query_parser.OR_OPERATOR);
|
---|
347 | }
|
---|
348 | }
|
---|
349 |
|
---|
350 |
|
---|
351 | public void cleanUp() {
|
---|
352 | super.cleanUp();
|
---|
353 | try {
|
---|
354 | if(reader != null) {
|
---|
355 | reader.close();
|
---|
356 | // Closes files associated with this index. Also saves any new deletions to disk.
|
---|
357 | // No other methods should be called after this has been called.
|
---|
358 | }
|
---|
359 |
|
---|
360 | } catch (IOException exception) {
|
---|
361 | exception.printStackTrace();
|
---|
362 | }
|
---|
363 | }
|
---|
364 |
|
---|
365 |
|
---|
366 | protected Query parseQuery(IndexReader reader, QueryParser query_parser, String query_string, String fuzziness)
|
---|
367 | throws java.io.IOException, org.apache.lucene.queryparser.classic.ParseException
|
---|
368 | {
|
---|
369 | // Split query string into the search terms and the filter terms
|
---|
370 | // * The first +(...) term contains the search terms so count
|
---|
371 | // up '(' and stop when we finish matching ')'
|
---|
372 | int offset = 0;
|
---|
373 | int paren_count = 0;
|
---|
374 | boolean seen_paren = false;
|
---|
375 | while (offset < query_string.length() && (!seen_paren || paren_count > 0)) {
|
---|
376 | if (query_string.charAt(offset) == '(') {
|
---|
377 | paren_count++;
|
---|
378 | seen_paren = true;
|
---|
379 | }
|
---|
380 | if (query_string.charAt(offset) == ')') {
|
---|
381 | paren_count--;
|
---|
382 | }
|
---|
383 | offset++;
|
---|
384 | }
|
---|
385 | String query_prefix = query_string.substring(0, offset);
|
---|
386 | String query_suffix = query_string.substring(offset);
|
---|
387 |
|
---|
388 | ///ystem.err.println("Prefix: " + query_prefix);
|
---|
389 | ///ystem.err.println("Suffix: " + query_suffix);
|
---|
390 |
|
---|
391 | Query query = query_parser.parse(query_prefix);
|
---|
392 | query = query.rewrite(reader);
|
---|
393 |
|
---|
394 | // If this is a fuzzy search, then we need to add the fuzzy
|
---|
395 | // flag to each of the query terms
|
---|
396 | if (fuzziness != null && query.toString().length() > 0) {
|
---|
397 |
|
---|
398 | // Revert the query to a string
|
---|
399 | System.err.println("Rewritten query: " + query.toString());
|
---|
400 | // Search through the string for TX:<term> query terms
|
---|
401 | // and append the ~ operator. Note that this search will
|
---|
402 | // not change phrase searches (TX:"<term> <term>") as
|
---|
403 | // fuzzy searching is not possible for these entries.
|
---|
404 | // Yahoo! Time for a state machine!
|
---|
405 | StringBuffer mutable_query_string = new StringBuffer(query.toString());
|
---|
406 | int o = 0; // Offset
|
---|
407 | // 0 = BASE, 1 = SEEN_T, 2 = SEEN_TX, 3 = SEEN_TX:
|
---|
408 | int s = 0; // State
|
---|
409 | while(o < mutable_query_string.length()) {
|
---|
410 | char c = mutable_query_string.charAt(o);
|
---|
411 | if (s == 0 && c == TEXTFIELD.charAt(0)) {
|
---|
412 | ///ystem.err.println("Found T!");
|
---|
413 | s = 1;
|
---|
414 | }
|
---|
415 | else if (s == 1) {
|
---|
416 | if (c == TEXTFIELD.charAt(1)) {
|
---|
417 | ///ystem.err.println("Found X!");
|
---|
418 | s = 2;
|
---|
419 | }
|
---|
420 | else {
|
---|
421 | s = 0; // Reset
|
---|
422 | }
|
---|
423 | }
|
---|
424 | else if (s == 2) {
|
---|
425 | if (c == ':') {
|
---|
426 | ///ystem.err.println("Found TX:!");
|
---|
427 | s = 3;
|
---|
428 | }
|
---|
429 | else {
|
---|
430 | s = 0; // Reset
|
---|
431 | }
|
---|
432 | }
|
---|
433 | else if (s == 3) {
|
---|
434 | // Don't process phrases
|
---|
435 | if (c == '"') {
|
---|
436 | ///ystem.err.println("Stupid phrase...");
|
---|
437 | s = 0; // Reset
|
---|
438 | }
|
---|
439 | // Found the end of the term... add the
|
---|
440 | // fuzzy search indicator
|
---|
441 | // Nor outside the scope of parentheses
|
---|
442 | else if (Character.isWhitespace(c) || c == ')') {
|
---|
443 | ///ystem.err.println("Yahoo! Found fuzzy term.");
|
---|
444 | mutable_query_string.insert(o, '~' + fuzziness);
|
---|
445 | o++;
|
---|
446 | s = 0; // Reset
|
---|
447 | }
|
---|
448 | }
|
---|
449 | o++;
|
---|
450 | }
|
---|
451 | // If we were in the state of looking for the end of a
|
---|
452 | // term - then we just found it!
|
---|
453 | if (s == 3) {
|
---|
454 |
|
---|
455 | mutable_query_string.append('~' + fuzziness);
|
---|
456 | }
|
---|
457 | // Reparse the query
|
---|
458 | ///ystem.err.println("Fuzzy query: " + mutable_query_string.toString() + query_suffix);
|
---|
459 | query = query_parser.parse(mutable_query_string.toString() + query_suffix);
|
---|
460 | }
|
---|
461 | else {
|
---|
462 | query = query_parser.parse(query_prefix + query_suffix);
|
---|
463 | }
|
---|
464 |
|
---|
465 | return query;
|
---|
466 | }
|
---|
467 |
|
---|
468 | // If you're dealing with a BooleanQuery, they need to be recursively rewritten
|
---|
469 | // as they can contain queries with wildcards (WildcardQuery|PrefixQuery subclasses of MultiTermQuery)
|
---|
470 | // e.g. season* farm
|
---|
471 | // If MultiTermQuery, then expand here. e.g. WildcardQuerys like season*.
|
---|
472 | // DON'T call this method from inside parseQuery() (in place of its query.rewrite()), because then wildcard
|
---|
473 | // queries like season* won't contain Terms (extractTerms() will be empty) since the ConstantScoreQuerys
|
---|
474 | // that a WildcardQuery gets rewritten to here will contain Filters in place of Terms.
|
---|
475 | // Call this method from runQuery() after it calls parseQuery().
|
---|
476 | // Now searches like these will work
|
---|
477 | // season* farm
|
---|
478 | // season* farm*
|
---|
479 | // and not just searches like the following which already used to work:
|
---|
480 | // season*
|
---|
481 | // snail farm
|
---|
482 | // Idea for the solution of recursively processing a BooleanQuery came from inspecting source code to BooleanQuery.java
|
---|
483 | // https://github.com/apache/lucene-solr/blob/master/lucene/core/src/java/org/apache/lucene/search/BooleanQuery.java
|
---|
484 | // which also does a recursive rewrite. Unfortunately, the existing BooleanQuery does not handle MultiTermQuery
|
---|
485 | // subcomponents.
|
---|
486 | protected Query recursivelyRewriteQuery(Query orig_query, IndexReader reader, LuceneQueryResult lucene_query_result) throws java.io.IOException
|
---|
487 | {
|
---|
488 | //Query query = orig_query.rewrite(reader);
|
---|
489 | Query query = orig_query;
|
---|
490 |
|
---|
491 | if(orig_query instanceof BooleanQuery) {
|
---|
492 | BooleanQuery booleanQuery = (BooleanQuery)orig_query;
|
---|
493 | List<BooleanClause> clauses = booleanQuery.clauses();
|
---|
494 | for (BooleanClause clause : clauses) {
|
---|
495 | Query subQuery = clause.getQuery();
|
---|
496 | subQuery = recursivelyRewriteQuery(subQuery, reader, lucene_query_result);
|
---|
497 | clause.setQuery(subQuery);
|
---|
498 | }
|
---|
499 | }
|
---|
500 |
|
---|
501 | // GS2's LuceneWrapper uses lucene-2.3.2. GS3's LuceneWrapper3 works with lucene-3.3.0.
|
---|
502 | // This change in lucene core library for GS3 (present since after version 2.4.1) had the
|
---|
503 | // side-effect that searching on "econom*" didn't display what terms it was searching for,
|
---|
504 | // whereas it had done so in GS2.
|
---|
505 |
|
---|
506 | // The details of this problem and its current solution are explained in the ticket
|
---|
507 | // http://trac.greenstone.org/ticket/845
|
---|
508 |
|
---|
509 | // We need to change the settings for the rewriteMethod in order to get searches on wildcards
|
---|
510 | // to produce search terms again when the query gets rewritten.
|
---|
511 |
|
---|
512 | // We try, in order:
|
---|
513 | // 1. RewriteMethod set to BooleanQuery, to get it working as in GS2 which uses lucene-2.3.2
|
---|
514 | // it will expand wildcard searches to its terms when searching at both section AND doc level.
|
---|
515 | // If that throws a TooManyClauses exception (like when searching for "a*" over lucene demo collection)
|
---|
516 | // 2. Then try a custom rewriteMethod which sets termCountCutoff=350 and docCountPercent cutoff=0.1%
|
---|
517 | // If that throws a TooManyClauses exception (could perhaps happen if the collection has a huge number of docs
|
---|
518 | // 3. Then try the default apache rewriteMethod with its optimum defaults of
|
---|
519 | // termCountCutoff=350 and docCountPercent cutoff=0.1%
|
---|
520 | // See http://lucene.apache.org/core/3_6_1/api/core/org/apache/lucene/search/MultiTermQuery.html
|
---|
521 |
|
---|
522 | //System.err.println("@@@@ query class name: " + orig_query.getClass());
|
---|
523 | //System.err.println("@@@@ QUERY: " + orig_query);
|
---|
524 |
|
---|
525 | if(orig_query instanceof MultiTermQuery) {
|
---|
526 | MultiTermQuery multiTermQuery = (MultiTermQuery)orig_query;
|
---|
527 | multiTermQuery.setRewriteMethod(MultiTermQuery.CONSTANT_SCORE_BOOLEAN_QUERY_REWRITE);
|
---|
528 | // less CPU intensive than MultiTermQuery.SCORING_BOOLEAN_QUERY_REWRITE)
|
---|
529 | }
|
---|
530 |
|
---|
531 | try {
|
---|
532 | query = orig_query.rewrite(reader);
|
---|
533 | }
|
---|
534 | catch(BooleanQuery.TooManyClauses clauseException) {
|
---|
535 | // Example test case: try searching the lucene demo collection for "a*"
|
---|
536 | // and you'll hit this exception
|
---|
537 |
|
---|
538 | lucene_query_result.setError(LuceneQueryResult.TOO_MANY_CLAUSES_ERROR);
|
---|
539 |
|
---|
540 | if(query instanceof MultiTermQuery) {
|
---|
541 |
|
---|
542 | // CustomRewriteMethod: setting the docCountPercent cutoff to a custom 100%.
|
---|
543 | // This will at least expand the query to its terms when searching with wildcards at section-level
|
---|
544 | // (though it doesn't seem to work for doc-level searches, no matter what the cutoffs are set to).
|
---|
545 |
|
---|
546 | MultiTermQuery.ConstantScoreAutoRewrite customRewriteMethod = new MultiTermQuery.ConstantScoreAutoRewrite();
|
---|
547 | customRewriteMethod.setDocCountPercent(100.0);
|
---|
548 | customRewriteMethod.setTermCountCutoff(350); // same as default
|
---|
549 |
|
---|
550 | MultiTermQuery multiTermQuery = (MultiTermQuery)query;
|
---|
551 | multiTermQuery.setRewriteMethod(customRewriteMethod);
|
---|
552 | try {
|
---|
553 | query = query.rewrite(reader);
|
---|
554 | }
|
---|
555 | catch(BooleanQuery.TooManyClauses clauseExceptionAgain) {
|
---|
556 |
|
---|
557 | // do what the code originally did: use the default rewriteMethod which
|
---|
558 | // uses a default docCountPercent=0.1 (%) and termCountCutoff=350
|
---|
559 |
|
---|
560 | multiTermQuery.setRewriteMethod(MultiTermQuery.CONSTANT_SCORE_AUTO_REWRITE_DEFAULT);
|
---|
561 | query = query.rewrite(reader);
|
---|
562 | }
|
---|
563 | }
|
---|
564 | }
|
---|
565 |
|
---|
566 | // BooleanQuery.java recurses rewriting any query until it is identical before and after rewrite,
|
---|
567 | // see reference to "recursively rewrite" in
|
---|
568 | // https://github.com/apache/lucene-solr/blob/master/lucene/core/src/java/org/apache/lucene/search/BooleanQuery.java
|
---|
569 | if(orig_query == query) {
|
---|
570 | return query;
|
---|
571 | } else {
|
---|
572 | return recursivelyRewriteQuery(query, reader, lucene_query_result);
|
---|
573 | }
|
---|
574 | }
|
---|
575 |
|
---|
576 | protected Filter parseFilterString(String filter_string)
|
---|
577 | {
|
---|
578 | Filter result = null;
|
---|
579 | Pattern pattern = Pattern.compile("\\s*\\+(\\w+)\\:([\\{\\[])(\\d+)\\s+TO\\s+(\\d+)([\\}\\]])\\s*");
|
---|
580 | Matcher matcher = pattern.matcher(filter_string);
|
---|
581 | if (matcher.matches()) {
|
---|
582 | String field_name = matcher.group(1);
|
---|
583 | boolean include_lower = matcher.group(2).equals("[");
|
---|
584 | BytesRef lower_term = new BytesRef(matcher.group(3));
|
---|
585 | BytesRef upper_term = new BytesRef(matcher.group(4));
|
---|
586 | boolean include_upper = matcher.group(5).equals("]");
|
---|
587 | result = new TermRangeFilter(field_name, lower_term, upper_term, include_lower, include_upper);
|
---|
588 | }
|
---|
589 | else {
|
---|
590 | System.err.println("Error: Could not understand filter string \"" + filter_string + "\"");
|
---|
591 | }
|
---|
592 | return result;
|
---|
593 | }
|
---|
594 |
|
---|
595 |
|
---|
596 | /** command line program and auxiliary methods */
|
---|
597 |
|
---|
598 | // Fairly self-explanatory I should hope
|
---|
599 | static protected boolean query_result_caching_enabled = false;
|
---|
600 |
|
---|
601 |
|
---|
602 | static public void main (String args[])
|
---|
603 | {
|
---|
604 | if (args.length == 0) {
|
---|
605 | System.out.println("Usage: org.greenstone.LuceneWrapper4.GS2LuceneQuery <index directory> [-fuzziness value] [-filter filter_string] [-sort sort_field] [-reverse_sort][-dco AND|OR] [-startresults number -endresults number] [query]");
|
---|
606 | return;
|
---|
607 | }
|
---|
608 |
|
---|
609 | try {
|
---|
610 | String index_directory = args[0];
|
---|
611 |
|
---|
612 | GS2LuceneQuery queryer = new GS2LuceneQuery();
|
---|
613 | queryer.setIndexDir(index_directory);
|
---|
614 |
|
---|
615 | // Prepare the index cache directory, if query result caching is enabled
|
---|
616 | if (query_result_caching_enabled) {
|
---|
617 | // Make the index cache directory if it doesn't already exist
|
---|
618 | File index_cache_directory = new File(index_directory, "cache");
|
---|
619 | if (!index_cache_directory.exists()) {
|
---|
620 | index_cache_directory.mkdir();
|
---|
621 | }
|
---|
622 |
|
---|
623 | // Disable caching if the index cache directory isn't available
|
---|
624 | if (!index_cache_directory.exists() || !index_cache_directory.isDirectory()) {
|
---|
625 | query_result_caching_enabled = false;
|
---|
626 | }
|
---|
627 | }
|
---|
628 |
|
---|
629 | String query_string = null;
|
---|
630 |
|
---|
631 | // Parse the command-line arguments
|
---|
632 | for (int i = 1; i < args.length; i++) {
|
---|
633 | if (args[i].equals("-sort")) {
|
---|
634 | i++;
|
---|
635 | queryer.setSortField(args[i]);
|
---|
636 | }
|
---|
637 | else if (args[i].equals("-reverse_sort")) {
|
---|
638 | queryer.setReverseSort(true);
|
---|
639 | }
|
---|
640 | else if (args[i].equals("-filter")) {
|
---|
641 | i++;
|
---|
642 | queryer.setFilterString(args[i]);
|
---|
643 | }
|
---|
644 | else if (args[i].equals("-dco")) {
|
---|
645 | i++;
|
---|
646 | queryer.setDefaultConjunctionOperator(args[i]);
|
---|
647 | }
|
---|
648 | else if (args[i].equals("-fuzziness")) {
|
---|
649 | i++;
|
---|
650 | queryer.setFuzziness(args[i]);
|
---|
651 | }
|
---|
652 | else if (args[i].equals("-startresults")) {
|
---|
653 | i++;
|
---|
654 | if (args[i].matches("\\d+")) {
|
---|
655 | queryer.setStartResults(Integer.parseInt(args[i]));
|
---|
656 | }
|
---|
657 | }
|
---|
658 | else if (args[i].equals("-endresults")) {
|
---|
659 | i++;
|
---|
660 | if (args[i].matches("\\d+")) {
|
---|
661 | queryer.setEndResults(Integer.parseInt(args[i]));
|
---|
662 | }
|
---|
663 | }
|
---|
664 | else {
|
---|
665 | query_string = args[i];
|
---|
666 | }
|
---|
667 | }
|
---|
668 |
|
---|
669 | if (!queryer.initialise()) {
|
---|
670 | queryer.cleanUp(); // will close reader object IF reader was instantiated
|
---|
671 | return;
|
---|
672 | }
|
---|
673 |
|
---|
674 | // The query string has been specified as a command-line argument
|
---|
675 | if (query_string != null) {
|
---|
676 | runQueryCaching(index_directory, queryer, query_string);
|
---|
677 | }
|
---|
678 |
|
---|
679 | // Read queries from STDIN
|
---|
680 | else {
|
---|
681 | BufferedReader in = new BufferedReader(new InputStreamReader(System.in, "UTF-8"));
|
---|
682 | while (true) {
|
---|
683 | // Read the query from STDIN
|
---|
684 | query_string = in.readLine();
|
---|
685 | if (query_string == null || query_string.length() == -1) {
|
---|
686 | break;
|
---|
687 | }
|
---|
688 |
|
---|
689 | runQueryCaching(index_directory, queryer, query_string);
|
---|
690 |
|
---|
691 | }
|
---|
692 | }
|
---|
693 | queryer.cleanUp();
|
---|
694 | }
|
---|
695 | catch (IOException exception) {
|
---|
696 | exception.printStackTrace();
|
---|
697 | }
|
---|
698 | }
|
---|
699 |
|
---|
700 | protected static void runQueryCaching(String index_directory, GS2LuceneQuery queryer, String query_string)
|
---|
701 | throws IOException
|
---|
702 | {
|
---|
703 | StringBuffer query_results_xml = new StringBuffer();
|
---|
704 |
|
---|
705 | // Check if this query result has been cached from a previous search (if it's enabled)
|
---|
706 | File query_result_cache_file = null;
|
---|
707 | if (query_result_caching_enabled) {
|
---|
708 | // Generate the cache file name from the query options
|
---|
709 | String query_result_cache_file_name = query_string + "-";
|
---|
710 | String fuzziness = queryer.getFuzziness();
|
---|
711 | query_result_cache_file_name += ((fuzziness != null) ? fuzziness : "") + "-";
|
---|
712 | String filter_string = queryer.getFilterString();
|
---|
713 | query_result_cache_file_name += ((filter_string != null) ? filter_string : "") + "-";
|
---|
714 | String sort_string = queryer.getSortField();
|
---|
715 | query_result_cache_file_name += ((sort_string != null) ? sort_string : "") + "-";
|
---|
716 | String reverse_sort_string = (queryer.getReverseSort() ? "1" : "0");
|
---|
717 | query_result_cache_file_name += reverse_sort_string + "-";
|
---|
718 | String default_conjunction_operator = queryer.getDefaultConjunctionOperator();
|
---|
719 | query_result_cache_file_name += default_conjunction_operator + "-";
|
---|
720 | int start_results = queryer.getStartResults();
|
---|
721 | int end_results = queryer.getEndResults();
|
---|
722 | query_result_cache_file_name += start_results + "-" + end_results;
|
---|
723 | query_result_cache_file_name = fileSafe(query_result_cache_file_name);
|
---|
724 |
|
---|
725 | // If the query result cache file exists, just return its contents and we're done
|
---|
726 | File index_cache_directory = new File(index_directory, "cache");
|
---|
727 | query_result_cache_file = new File(index_cache_directory, query_result_cache_file_name);
|
---|
728 | if (query_result_cache_file.exists() && query_result_cache_file.isFile()) {
|
---|
729 | FileInputStream fis = new FileInputStream(query_result_cache_file);
|
---|
730 | InputStreamReader isr = new InputStreamReader(fis, "UTF-8");
|
---|
731 | BufferedReader buffered_reader = new BufferedReader(isr);
|
---|
732 | String line = "";
|
---|
733 | while ((line = buffered_reader.readLine()) != null) {
|
---|
734 | query_results_xml.append(line + "\n");
|
---|
735 | }
|
---|
736 | String query_results_xml_string = query_results_xml.toString();
|
---|
737 | query_results_xml_string = query_results_xml_string.replaceFirst("cached=\"false\"", "cached=\"true\"");
|
---|
738 |
|
---|
739 | utf8out.print(query_results_xml_string);
|
---|
740 | utf8out.flush();
|
---|
741 |
|
---|
742 | return;
|
---|
743 | }
|
---|
744 | }
|
---|
745 |
|
---|
746 | // not cached
|
---|
747 | query_results_xml.append("<ResultSet cached=\"false\">\n");
|
---|
748 | query_results_xml.append("<QueryString>" + LuceneQueryResult.xmlSafe(query_string) + "</QueryString>\n");
|
---|
749 | Filter filter = queryer.getFilter();
|
---|
750 | if (filter != null) {
|
---|
751 | query_results_xml.append("<FilterString>" + filter.toString() + "</FilterString>\n");
|
---|
752 | }
|
---|
753 |
|
---|
754 | LuceneQueryResult query_result = queryer.runQuery(query_string);
|
---|
755 | if (query_result == null) {
|
---|
756 | System.err.println("Couldn't run the query");
|
---|
757 | return;
|
---|
758 | }
|
---|
759 |
|
---|
760 | if (query_result.getError() != LuceneQueryResult.NO_ERROR) {
|
---|
761 | query_results_xml.append("<Error type=\""+query_result.getErrorString()+"\" />\n");
|
---|
762 | } else {
|
---|
763 | query_results_xml.append(query_result.getXMLString());
|
---|
764 | }
|
---|
765 | query_results_xml.append("</ResultSet>\n");
|
---|
766 |
|
---|
767 | utf8out.print(query_results_xml);
|
---|
768 | utf8out.flush();
|
---|
769 |
|
---|
770 | // Cache this query result, if desired
|
---|
771 | if (query_result_caching_enabled) {
|
---|
772 | // Catch any exceptions thrown trying to write the query result cache file and warn about them, but don't
|
---|
773 | // bother with the full stack trace. It won't affect the functionality if we can't write some cache
|
---|
774 | // files, it will just affect the speed of subsequent requests.
|
---|
775 | // Example exceptions are "permission denied" errors, or "filename too long" errors (the filter string
|
---|
776 | // can get very long in some collections)
|
---|
777 | try
|
---|
778 | {
|
---|
779 | FileWriter query_result_cache_file_writer = new FileWriter(query_result_cache_file);
|
---|
780 | query_result_cache_file_writer.write(query_results_xml.toString());
|
---|
781 | query_result_cache_file_writer.close();
|
---|
782 | }
|
---|
783 | catch (Exception exception)
|
---|
784 | {
|
---|
785 | System.err.println("Warning: Exception occurred trying to write query result cache file (" + exception + ")");
|
---|
786 | }
|
---|
787 | }
|
---|
788 | }
|
---|
789 |
|
---|
790 | protected static String fileSafe(String text)
|
---|
791 | {
|
---|
792 | StringBuffer file_safe_text = new StringBuffer();
|
---|
793 | for (int i = 0; i < text.length(); i++) {
|
---|
794 | char character = text.charAt(i);
|
---|
795 | if ((character >= 'A' && character <= 'Z') || (character >= 'a' && character <= 'z') || (character >= '0' && character <= '9') || character == '-') {
|
---|
796 | file_safe_text.append(character);
|
---|
797 | }
|
---|
798 | else {
|
---|
799 | file_safe_text.append('%');
|
---|
800 | file_safe_text.append((int) character);
|
---|
801 | }
|
---|
802 | }
|
---|
803 | return file_safe_text.toString();
|
---|
804 | }
|
---|
805 |
|
---|
806 |
|
---|
807 | }
|
---|
808 |
|
---|
809 |
|
---|