source: indexers/trunk/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneQuery.java@ 16912

Last change on this file since 16912 was 16912, checked in by mdewsnip, 16 years ago

Changes made by Richard Managh at DL Consulting Ltd for returning document-level term frequency totals, slightly modified to work with the latest Greenstone.
  • Property svn:keywords set to Author Date Id Revision
File size: 21.1 KB
Line 
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 *********************************************************************/
26package org.greenstone.LuceneWrapper;
27
28
29import java.io.*;
30import java.util.*;
31import java.util.regex.*;
32
33import org.apache.lucene.analysis.Analyzer;
34import org.apache.lucene.analysis.standard.StandardAnalyzer;
35import org.apache.lucene.document.Document;
36import org.apache.lucene.index.IndexReader;
37import org.apache.lucene.index.Term;
38import org.apache.lucene.index.TermDocs;
39import org.apache.lucene.queryParser.ParseException;
40import org.apache.lucene.queryParser.QueryParser;
41import org.apache.lucene.search.BooleanQuery.TooManyClauses;
42import org.apache.lucene.search.Filter;
43import org.apache.lucene.search.Hit;
44import org.apache.lucene.search.Hits;
45import org.apache.lucene.search.IndexSearcher;
46import org.apache.lucene.search.Query;
47import org.apache.lucene.search.RangeFilter;
48import org.apache.lucene.search.Searcher;
49import org.apache.lucene.search.ScoreDoc;
50import org.apache.lucene.search.Sort;
51import org.apache.lucene.search.TopFieldDocs;
52
53
54public class GS2LuceneQuery
55{
56
57
58 static private String TEXTFIELD = "TX";
59
60 // Use the standard set of English stop words by default
61 static private String[] stop_words = StandardAnalyzer.STOP_WORDS;
62
63 private String full_indexdir="";
64 private String default_conjunction_operator = "OR";
65 private String fuzziness = null;
66 private String sort_field = null;
67 private Sort sorter=new Sort();
68 private String filter_string = null;
69 private Filter filter = null;
70 private int start_results=1;
71 private int end_results=Integer.MAX_VALUE;
72
73 private QueryParser query_parser = null;
74 private QueryParser query_parser_no_stop_words = null;
75 private Searcher searcher = null;
76 private IndexReader reader = null;
77
78 static private PrintWriter utf8out = null;
79
80 static
81 {
82 try {
83 OutputStreamWriter osw = new OutputStreamWriter(System.out, "UTF-8");
84 utf8out = new PrintWriter(osw, true);
85 }
86 catch (UnsupportedEncodingException e) {
87 System.out.println(e);
88 }
89 }
90
91
92 public GS2LuceneQuery() {
93
94 // Create one query parser with the standard set of stop words, and one with none
95
96 query_parser = new QueryParser(TEXTFIELD, new StandardAnalyzer(stop_words));
97 query_parser_no_stop_words = new QueryParser(TEXTFIELD, new StandardAnalyzer(new String[] { }));
98 }
99
100
101 public boolean initialise() {
102
103 if (full_indexdir==null || full_indexdir.length()==-1){
104 utf8out.println("Index directory is not indicated ");
105 utf8out.flush();
106 return false;
107 }
108 try {
109 searcher = new IndexSearcher(full_indexdir);
110 reader = ((IndexSearcher) searcher).getIndexReader();
111
112 }
113 catch (IOException exception) {
114 exception.printStackTrace();
115 return false;
116 }
117 return true;
118
119 }
120
121 public LuceneQueryResult runQuery(String query_string) {
122
123 if (query_string == null || query_string.equals("")) {
124 utf8out.println("The query word is not indicated ");
125 utf8out.flush();
126 return null;
127 }
128
129 LuceneQueryResult lucene_query_result=new LuceneQueryResult();
130 lucene_query_result.clear();
131
132 try {
133 Query query_including_stop_words = query_parser_no_stop_words.parse(query_string);
134 query_including_stop_words = query_including_stop_words.rewrite(reader);
135
136 Query query = parseQuery(reader, query_parser, query_string, fuzziness);
137 query = query.rewrite(reader);
138
139 // Get the list of expanded query terms and their frequencies
140 // num docs matching, and total frequency
141 HashSet terms = new HashSet();
142 query.extractTerms(terms);
143
144 HashMap doc_term_freq_map = new HashMap();
145
146 Iterator iter = terms.iterator();
147 while (iter.hasNext()) {
148
149 Term term = (Term) iter.next();
150
151 // Get the term frequency over all the documents
152 TermDocs term_docs = reader.termDocs(term);
153 int term_freq = 0;
154 int match_docs = 0;
155 while (term_docs.next())
156 {
157 if (term_docs.freq() != 0)
158 {
159 term_freq += term_docs.freq();
160 match_docs++;
161
162 // Calculate the document-level term frequency as well
163 Integer lucene_doc_num_obj = new Integer(term_docs.doc());
164 int doc_term_freq = 0;
165 if (doc_term_freq_map.containsKey(lucene_doc_num_obj))
166 {
167 doc_term_freq = ((Integer) doc_term_freq_map.get(lucene_doc_num_obj)).intValue();
168 }
169 doc_term_freq += term_docs.freq();
170
171 doc_term_freq_map.put(lucene_doc_num_obj, new Integer(doc_term_freq));
172 }
173 }
174
175 // Create a term
176 lucene_query_result.addTerm(term.text(), term.field(), match_docs, term_freq);
177 }
178
179 // Get the list of stop words removed from the query
180 HashSet terms_including_stop_words = new HashSet();
181 query_including_stop_words.extractTerms(terms_including_stop_words);
182 Iterator terms_including_stop_words_iter = terms_including_stop_words.iterator();
183 while (terms_including_stop_words_iter.hasNext()) {
184 Term term = (Term) terms_including_stop_words_iter.next();
185 if (!terms.contains(term)) {
186 lucene_query_result.addStopWord(term.text());
187 }
188 }
189
190 // do the query
191 // Simple case for getting all the matching documents
192 if (end_results == Integer.MAX_VALUE) {
193 // Perform the query (filter and sorter may be null)
194 Hits hits = searcher.search(query, filter, sorter);
195 lucene_query_result.setTotalDocs(hits.length());
196
197 // Output the matching documents
198 lucene_query_result.setStartResults(start_results);
199 lucene_query_result.setEndResults(hits.length());
200
201 for (int i = start_results; i <= hits.length(); i++) {
202 int lucene_doc_num = hits.id(i - 1);
203 Document doc = hits.doc(i - 1);
204 int doc_term_freq = ((Integer) doc_term_freq_map.get(new Integer(lucene_doc_num))).intValue();
205 lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.score(i-1), doc_term_freq);
206 }
207 }
208
209 // Slightly more complicated case for returning a subset of the matching documents
210 else {
211 // Perform the query (filter may be null)
212 TopFieldDocs hits = searcher.search(query, filter, end_results, sorter);
213 lucene_query_result.setTotalDocs(hits.totalHits);
214
215 lucene_query_result.setStartResults(start_results);
216 lucene_query_result.setEndResults(end_results < hits.scoreDocs.length ? end_results: hits.scoreDocs.length);
217
218 // Output the matching documents
219 for (int i = start_results; (i <= hits.scoreDocs.length && i <= end_results); i++) {
220 int lucene_doc_num = hits.scoreDocs[i - 1].doc;
221 Document doc = reader.document(lucene_doc_num);
222 int doc_term_freq = ((Integer) doc_term_freq_map.get(new Integer(lucene_doc_num))).intValue();
223 lucene_query_result.addDoc(doc.get("nodeID").trim(), hits.scoreDocs[i-1].score, doc_term_freq);
224 }
225 }
226 }
227
228 catch (ParseException parse_exception) {
229 lucene_query_result.setError(LuceneQueryResult.PARSE_ERROR);
230 }
231 catch (TooManyClauses too_many_clauses_exception) {
232 lucene_query_result.setError(LuceneQueryResult.TOO_MANY_CLAUSES_ERROR);
233 }
234 catch (IOException exception) {
235 lucene_query_result.setError(LuceneQueryResult.IO_ERROR);
236 exception.printStackTrace();
237 }
238 catch (Exception exception) {
239 lucene_query_result.setError(LuceneQueryResult.OTHER_ERROR);
240 exception.printStackTrace();
241 }
242 return lucene_query_result;
243 }
244
245 public void setDefaultConjunctionOperator(String default_conjunction_operator) {
246 this.default_conjunction_operator = default_conjunction_operator.toUpperCase();
247 if (default_conjunction_operator.equals("AND")) {
248 query_parser.setDefaultOperator(query_parser.AND_OPERATOR);
249 query_parser_no_stop_words.setDefaultOperator(query_parser.AND_OPERATOR);
250 } else { // default is OR
251 query_parser.setDefaultOperator(query_parser.OR_OPERATOR);
252 query_parser_no_stop_words.setDefaultOperator(query_parser.OR_OPERATOR);
253 }
254 }
255
256 public String getDefaultConjunctionOperator() {
257 return this.default_conjunction_operator;
258 }
259
260 public void setEndResults(int end_results) {
261 this.end_results = end_results;
262 }
263 public int getEndResults() {
264 return this.end_results;
265 }
266
267 public void setFilterString(String filter_string) {
268 this.filter_string = filter_string;
269 this.filter = parseFilterString(filter_string);
270 }
271 public String getFilterString() {
272 return this.filter_string ;
273 }
274
275 public Filter getFilter() {
276 return this.filter;
277 }
278
279 public void setIndexDir(String full_indexdir) {
280 this.full_indexdir = full_indexdir;
281 }
282
283 public void setFuzziness(String fuzziness) {
284 this.fuzziness = fuzziness;
285 }
286 public String getFuzziness() {
287 return this.fuzziness;
288 }
289
290 public void setSortField(String sort_field) {
291 this.sort_field = sort_field;
292 if (sort_field == null) {
293 this.sorter = new Sort();
294 } else {
295 this.sorter = new Sort(sort_field);
296 }
297 }
298 public String getSortField() {
299 return this.sort_field;
300 }
301
302 public void setStartResults(int start_results) {
303 if (start_results < 1) {
304 start_results = 1;
305 }
306 this.start_results = start_results;
307 }
308 public int getStartResults() {
309 return this.start_results;
310 }
311
312 public void cleanUp() {
313 try {
314 if (searcher != null) {
315 searcher.close();
316 }
317 } catch (IOException exception) {
318 exception.printStackTrace();
319 }
320 }
321
322 private Query parseQuery(IndexReader reader, QueryParser query_parser, String query_string, String fuzziness)
323 throws java.io.IOException, org.apache.lucene.queryParser.ParseException
324 {
325 // Split query string into the search terms and the filter terms
326 // * The first +(...) term contains the search terms so count
327 // up '(' and stop when we finish matching ')'
328 int offset = 0;
329 int paren_count = 0;
330 boolean seen_paren = false;
331 while (offset < query_string.length() && (!seen_paren || paren_count > 0)) {
332 if (query_string.charAt(offset) == '(') {
333 paren_count++;
334 seen_paren = true;
335 }
336 if (query_string.charAt(offset) == ')') {
337 paren_count--;
338 }
339 offset++;
340 }
341 String query_prefix = query_string.substring(0, offset);
342 String query_suffix = query_string.substring(offset);
343
344 ///ystem.err.println("Prefix: " + query_prefix);
345 ///ystem.err.println("Suffix: " + query_suffix);
346
347 Query query = query_parser.parse(query_prefix);
348 query = query.rewrite(reader);
349
350 // If this is a fuzzy search, then we need to add the fuzzy
351 // flag to each of the query terms
352 if (fuzziness != null && query.toString().length() > 0) {
353
354 // Revert the query to a string
355 System.err.println("Rewritten query: " + query.toString());
356 // Search through the string for TX:<term> query terms
357 // and append the ~ operator. Note that this search will
358 // not change phrase searches (TX:"<term> <term>") as
359 // fuzzy searching is not possible for these entries.
360 // Yahoo! Time for a state machine!
361 StringBuffer mutable_query_string = new StringBuffer(query.toString());
362 int o = 0; // Offset
363 // 0 = BASE, 1 = SEEN_T, 2 = SEEN_TX, 3 = SEEN_TX:
364 int s = 0; // State
365 while(o < mutable_query_string.length()) {
366 char c = mutable_query_string.charAt(o);
367 if (s == 0 && c == TEXTFIELD.charAt(0)) {
368 ///ystem.err.println("Found T!");
369 s = 1;
370 }
371 else if (s == 1) {
372 if (c == TEXTFIELD.charAt(1)) {
373 ///ystem.err.println("Found X!");
374 s = 2;
375 }
376 else {
377 s = 0; // Reset
378 }
379 }
380 else if (s == 2) {
381 if (c == ':') {
382 ///ystem.err.println("Found TX:!");
383 s = 3;
384 }
385 else {
386 s = 0; // Reset
387 }
388 }
389 else if (s == 3) {
390 // Don't process phrases
391 if (c == '"') {
392 ///ystem.err.println("Stupid phrase...");
393 s = 0; // Reset
394 }
395 // Found the end of the term... add the
396 // fuzzy search indicator
397 // Nor outside the scope of parentheses
398 else if (Character.isWhitespace(c) || c == ')') {
399 ///ystem.err.println("Yahoo! Found fuzzy term.");
400 mutable_query_string.insert(o, '~' + fuzziness);
401 o++;
402 s = 0; // Reset
403 }
404 }
405 o++;
406 }
407 // If we were in the state of looking for the end of a
408 // term - then we just found it!
409 if (s == 3) {
410
411 mutable_query_string.append('~' + fuzziness);
412 }
413 // Reparse the query
414 ///ystem.err.println("Fuzzy query: " + mutable_query_string.toString() + query_suffix);
415 query = query_parser.parse(mutable_query_string.toString() + query_suffix);
416 }
417 else {
418 query = query_parser.parse(query_prefix + query_suffix);
419 }
420
421 return query;
422 }
423
424 private Filter parseFilterString(String filter_string)
425 {
426 Filter result = null;
427 Pattern pattern = Pattern.compile("\\s*\\+(\\w+)\\:([\\{\\[])(\\d+)\\s+TO\\s+(\\d+)([\\}\\]])\\s*");
428 Matcher matcher = pattern.matcher(filter_string);
429 if (matcher.matches()) {
430 String field_name = matcher.group(1);
431 boolean include_lower = matcher.group(2).equals("[");
432 String lower_term = matcher.group(3);
433 String upper_term = matcher.group(4);
434 boolean include_upper = matcher.group(5).equals("]");
435 result = new RangeFilter(field_name, lower_term, upper_term, include_lower, include_upper);
436 }
437 else {
438 System.err.println("Error: Could not understand filter string \"" + filter_string + "\"");
439 }
440 return result;
441 }
442
443
444 protected void finalize() throws Throwable
445 {
446 try {
447 utf8out.flush();
448 } finally {
449 super.finalize();
450 }
451 }
452
453
454 /** command line program and auxiliary methods */
455
456 // Fairly self-explanatory I should hope
457 static private boolean query_result_caching_enabled = false;
458
459
460 static public void main (String args[])
461 {
462
463
464 if (args.length == 0) {
465 System.out.println("Usage: GS2LuceneQuery <index directory> [-fuzziness value] [-filter filter_string] [-sort sort_field] [-dco AND|OR] [-startresults number -endresults number] [query]");
466 return;
467 }
468
469 try {
470 String index_directory = args[0];
471
472 GS2LuceneQuery queryer = new GS2LuceneQuery();
473 queryer.setIndexDir(index_directory);
474
475 // Prepare the index cache directory, if query result caching is enabled
476 if (query_result_caching_enabled) {
477 // Make the index cache directory if it doesn't already exist
478 File index_cache_directory = new File(index_directory, "cache");
479 if (!index_cache_directory.exists()) {
480 index_cache_directory.mkdir();
481 }
482
483 // Disable caching if the index cache directory isn't available
484 if (!index_cache_directory.exists() || !index_cache_directory.isDirectory()) {
485 query_result_caching_enabled = false;
486 }
487 }
488
489 String query_string = null;
490
491 // Parse the command-line arguments
492 for (int i = 1; i < args.length; i++) {
493 if (args[i].equals("-sort")) {
494 i++;
495 queryer.setSortField(args[i]);
496 }
497 else if (args[i].equals("-filter")) {
498 i++;
499 queryer.setFilterString(args[i]);
500 }
501 else if (args[i].equals("-dco")) {
502 i++;
503 queryer.setDefaultConjunctionOperator(args[i]);
504 }
505 else if (args[i].equals("-fuzziness")) {
506 i++;
507 queryer.setFuzziness(args[i]);
508 }
509 else if (args[i].equals("-startresults")) {
510 i++;
511 if (args[i].matches("\\d+")) {
512 queryer.setStartResults(Integer.parseInt(args[i]));
513 }
514 }
515 else if (args[i].equals("-endresults")) {
516 i++;
517 if (args[i].matches("\\d+")) {
518 queryer.setEndResults(Integer.parseInt(args[i]));
519 }
520 }
521 else {
522 query_string = args[i];
523 }
524 }
525
526 if (!queryer.initialise()) {
527 return;
528 }
529
530 // The query string has been specified as a command-line argument
531 if (query_string != null) {
532 runQueryCaching(index_directory, queryer, query_string);
533 }
534
535 // Read queries from STDIN
536 else {
537 BufferedReader in = new BufferedReader(new InputStreamReader(System.in, "UTF-8"));
538 while (true) {
539 // Read the query from STDIN
540 query_string = in.readLine();
541 if (query_string == null || query_string.length() == -1) {
542 break;
543 }
544
545 runQueryCaching(index_directory, queryer, query_string);
546
547 }
548 }
549 queryer.cleanUp();
550 }
551 catch (IOException exception) {
552 exception.printStackTrace();
553 }
554 }
555
556 private static void runQueryCaching(String index_directory, GS2LuceneQuery queryer, String query_string)
557 throws IOException
558 {
559 StringBuffer query_results_xml = new StringBuffer();
560
561 // Check if this query result has been cached from a previous search (if it's enabled)
562 File query_result_cache_file = null;
563 if (query_result_caching_enabled) {
564 // Generate the cache file name from the query options
565 String query_result_cache_file_name = query_string + "-";
566 String fuzziness = queryer.getFuzziness();
567 query_result_cache_file_name += ((fuzziness != null) ? fuzziness : "") + "-";
568 String filter_string = queryer.getFilterString();
569 query_result_cache_file_name += ((filter_string != null) ? filter_string : "") + "-";
570 String sort_string = queryer.getSortField();
571 query_result_cache_file_name += ((sort_string != null) ? sort_string : "") + "-";
572 String default_conjunction_operator = queryer.getDefaultConjunctionOperator();
573 query_result_cache_file_name += default_conjunction_operator + "-";
574 int start_results = queryer.getStartResults();
575 int end_results = queryer.getEndResults();
576 query_result_cache_file_name += start_results + "-" + end_results;
577 query_result_cache_file_name = fileSafe(query_result_cache_file_name);
578
579 // If the query result cache file exists, just return its contents and we're done
580 File index_cache_directory = new File(index_directory, "cache");
581 query_result_cache_file = new File(index_cache_directory, query_result_cache_file_name);
582 if (query_result_cache_file.exists() && query_result_cache_file.isFile()) {
583 FileInputStream fis = new FileInputStream(query_result_cache_file);
584 InputStreamReader isr = new InputStreamReader(fis, "UTF-8");
585 BufferedReader buffered_reader = new BufferedReader(isr);
586 String line = "";
587 while ((line = buffered_reader.readLine()) != null) {
588 query_results_xml.append(line + "\n");
589 }
590 String query_results_xml_string = query_results_xml.toString();
591 query_results_xml_string = query_results_xml_string.replaceFirst("cached=\"false\"", "cached=\"true\"");
592
593 utf8out.print(query_results_xml_string);
594 utf8out.flush();
595
596 return;
597 }
598 }
599
600 // not cached
601 query_results_xml.append("<ResultSet cached=\"false\">\n");
602 query_results_xml.append("<QueryString>" + LuceneQueryResult.xmlSafe(query_string) + "</QueryString>\n");
603 Filter filter = queryer.getFilter();
604 if (filter != null) {
605 query_results_xml.append("<FilterString>" + filter.toString() + "</FilterString>\n");
606 }
607
608 LuceneQueryResult query_result = queryer.runQuery(query_string);
609 if (query_result == null) {
610 System.err.println("Couldn't run the query");
611 return;
612 }
613
614 if (query_result.getError() != LuceneQueryResult.NO_ERROR) {
615 query_results_xml.append("<Error type=\""+query_result.getErrorString()+"\" />\n");
616 } else {
617 query_results_xml.append(query_result.getXMLString());
618 }
619 query_results_xml.append("</ResultSet>\n");
620
621 utf8out.print(query_results_xml);
622 utf8out.flush();
623
624 try {
625 /*
626 Writer output = null;
627 File file = new File("/tmp/lucenequery.txt");
628 output = new BufferedWriter(new FileWriter(file,"UTF-8"));
629 output.write(query_results_xml.toString());
630 output.close();
631 */
632
633 FileOutputStream fos = new FileOutputStream("/tmp/lucenequery.txt");
634
635 OutputStreamWriter osw2 = new OutputStreamWriter(fos, "UTF-8");
636
637 osw2.write("Query string = " + query_string + "\n");
638 osw2.write(query_results_xml.toString());
639 osw2.close();
640 }
641 catch (Exception e) {
642 e.printStackTrace();
643 }
644
645
646
647 // Cache this query result, if desired
648 if (query_result_caching_enabled) {
649 FileWriter query_result_cache_file_writer = new FileWriter(query_result_cache_file);
650 query_result_cache_file_writer.write(query_results_xml.toString());
651 query_result_cache_file_writer.close();
652 }
653 }
654
655 private static String fileSafe(String text)
656 {
657 StringBuffer file_safe_text = new StringBuffer();
658 for (int i = 0; i < text.length(); i++) {
659 char character = text.charAt(i);
660 if ((character >= 'A' && character <= 'Z') || (character >= 'a' && character <= 'z') || (character >= '0' && character <= '9') || character == '-') {
661 file_safe_text.append(character);
662 }
663 else {
664 file_safe_text.append('%');
665 file_safe_text.append((int) character);
666 }
667 }
668 return file_safe_text.toString();
669 }
670
671
672}
673
674
Note: See TracBrowser for help on using the repository browser.