source: gs3-extensions/solr/trunk/src/src/java/org/greenstone/gsdl3/util/SolrQueryWrapper.java@ 29217

Last change on this file since 29217 was 29217, checked in by ak19, 10 years ago

Terms must be found in a sorlCore (sidx or didx) of THIS collection, not in the first solrCore in the CoreContainer (solr.xml), since the CoreContainer can contain core descriptors of other GS collections as welll.
  • Property svn:executable set to *
File size: 16.4 KB
Line 
1/**********************************************************************
2 *
3 * SolrQueryWrapper.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.gsdl3.util;
27
28import java.lang.reflect.Type;
29import java.net.URLDecoder;
30import java.util.ArrayList;
31import java.util.Collection;
32import java.util.HashMap;
33import java.util.Iterator;
34import java.util.List;
35import java.util.Set;
36import java.util.HashSet;
37
38import org.apache.log4j.Logger;
39import org.apache.solr.client.solrj.SolrQuery; // subclass of ModifiableSolrParams
40import org.apache.solr.client.solrj.SolrServer;
41import org.apache.solr.client.solrj.SolrServerException;
42import org.apache.solr.client.solrj.embedded.EmbeddedSolrServer;
43import org.apache.solr.client.solrj.response.QueryResponse;
44import org.apache.solr.client.solrj.response.TermsResponse;
45
46import org.apache.solr.core.CoreContainer;
47import org.apache.solr.core.SolrCore;
48
49import org.apache.solr.common.SolrDocument;
50import org.apache.solr.common.SolrDocumentList;
51import org.apache.solr.common.params.ModifiableSolrParams;
52import org.greenstone.LuceneWrapper4.SharedSoleneQuery;
53import org.greenstone.LuceneWrapper4.SharedSoleneQueryResult;
54
55import org.apache.lucene.search.Query; // Query, TermQuery, BooleanQuery, BooleanClause and more
56import org.apache.lucene.index.IndexReader;
57import org.apache.lucene.index.Term;
58import org.apache.solr.search.QParser;
59import org.apache.solr.search.SolrIndexSearcher;
60import org.apache.solr.request.LocalSolrQueryRequest;
61
62import com.google.gson.Gson;
63import com.google.gson.reflect.TypeToken;
64
65public class SolrQueryWrapper extends SharedSoleneQuery
66{
67 public static String SORT_ASCENDING = "asc";
68 public static String SORT_DESCENDING = "desc";
69 public static String SORT_BY_RANK = "score";
70 public static String SORT_BY_INDEX_ORDER = "_docid_";
71
72 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.util.SolrQueryWrapper.class.getName());
73 protected int max_docs = 100;
74 protected String sort_order = SORT_DESCENDING;
75 protected String sort_field = SORT_BY_RANK; // don't want null default for solr
76 protected ArrayList<String> _facets = new ArrayList<String>();
77 protected ArrayList<String> _facetQueries = new ArrayList<String>();
78 SolrServer solr_core = null;
79
80 String collection_core_name_prefix = null;
81
82 public SolrQueryWrapper()
83 {
84 super();
85 start_results = 0;
86 }
87
88 public void setMaxDocs(int max_docs)
89 {
90 this.max_docs = max_docs;
91 }
92
93 public void setSolrCore(SolrServer solr_core)
94 {
95 this.solr_core = solr_core;
96 }
97
98 public void setCollectionCoreNamePrefix(String colCoreNamePrefix) {
99 this.collection_core_name_prefix = colCoreNamePrefix;
100 }
101
102 // make sure its not null.
103 public void setSortField(String sort_field) {
104 if (sort_field != null) {
105 this.sort_field = sort_field;
106 }
107 }
108
109 public void setSortOrder(String order)
110 {
111 this.sort_order = order;
112 }
113 public void addFacet(String facet)
114 {
115 if (!_facets.contains(facet))
116 {
117 _facets.add(facet);
118 }
119 }
120
121 public void clearFacets()
122 {
123 _facets.clear();
124 }
125
126 public void addFacetQuery(String facetQuery)
127 {
128 if (!_facetQueries.contains(facetQuery))
129 {
130 _facetQueries.add(facetQuery);
131 }
132 }
133
134 public void clearFacetQueries()
135 {
136 _facetQueries.clear();
137 }
138
139 public boolean initialise()
140 {
141 if (solr_core == null)
142 {
143 utf8out.println("Solr Core not loaded in ");
144 utf8out.flush();
145 return false;
146 }
147 return true;
148 }
149
150
151 /** Extracts the query terms from the query string. The query string can be a boolean
152 * combination of the various search fields with their search terms or phrases
153 */
154 public Term[] getTerms(SolrQuery solrQuery, String query_string)
155 {
156 Term terms[] = null;
157
158 if(solr_core instanceof EmbeddedSolrServer) {
159 EmbeddedSolrServer solrServer = (EmbeddedSolrServer)solr_core;
160
161 CoreContainer coreContainer = solrServer.getCoreContainer();
162
163 Collection<SolrCore> solrCores = coreContainer.getCores();
164 if(!solrCores.isEmpty()) {
165 Iterator<SolrCore> coreIterator = solrCores.iterator();
166
167 // Just use the first core that matches the collection name, since the term
168 // frequency of any term is the same regardless of whether its didx or sidx core
169 boolean foundCore = false;
170 while(coreIterator.hasNext() && !foundCore) {
171 SolrCore solrCore = coreIterator.next();
172 if(!solrCore.getName().startsWith(this.collection_core_name_prefix)) {
173 //logger.error("### Skipping core not of this collection: " + solrCore.getName());
174 continue;
175 }
176
177 //logger.error("### Found core " + solrCore.getName() + " of this collection " + this.collection_core_name_prefix);
178 foundCore = true;
179
180 LocalSolrQueryRequest solrQueryRequest = new LocalSolrQueryRequest(solrCore, solrQuery);
181 Query parsedQuery = null;
182
183 try {
184
185 // get the qparser, default is LuceneQParserPlugin, which is called "lucene" see http://wiki.apache.org/solr/QueryParser
186 QParser qParser = QParser.getParser(query_string, "lucene", solrQueryRequest);
187 parsedQuery = qParser.getQuery();
188
189 // For PrefixQuery or WildCardQuery (a subclass of AutomatonQuery, incl RegexpQ),
190 // like ZZ:econom* and ZZ:*date/regex queries, Query.extractTerms() throws an Exception
191 // because it has not done the Query.rewrite() step yet. So do that manually for them.
192 // This still doesn't provide us with the terms that econom* or *date break down into.
193
194 //if(parsedQuery instanceof PrefixQuery || parsedQuery instanceof AutomatonQuery) {
195 // Should we just check superclass MultiTermQuery?
196 // Can be a BooleanQuery containing PrefixQuery/WildCardQuery among its clauses, so
197 // just test for * in the query_string to determine if we need to do a rewrite() or not
198 if(query_string.contains("*")) {
199 SolrIndexSearcher searcher = solrQueryRequest.getSearcher();
200 IndexReader indexReader = searcher.getIndexReader(); // returns a DirectoryReader
201 parsedQuery = parsedQuery.rewrite(indexReader); // gets rewritten to ConstantScoreQuery
202 }
203
204 //System.err.println("#### Query type was: " + parsedQuery.getClass());
205 //logger.error("#### Query type was: " + parsedQuery.getClass());
206
207 // extract the terms
208 Set<Term> extractedQueryTerms = new HashSet<Term>();
209 parsedQuery.extractTerms(extractedQueryTerms);
210
211 terms = new Term[extractedQueryTerms.size()];
212
213 Iterator<Term> termsIterator = extractedQueryTerms.iterator();
214 for(int i = 0; termsIterator.hasNext(); i++) {
215 Term term = termsIterator.next();
216 ///System.err.println("#### Found query term: " + term);
217 ///logger.error("#### Found query term: " + term);
218
219 terms[i] = term; //(term.field(), term.text());
220 }
221
222 } catch(Exception queryParseException) {
223 queryParseException.printStackTrace();
224 System.err.println("Exception when parsing query: " + queryParseException.getMessage());
225 System.err.println("#### Query type was: " + parsedQuery.getClass());
226 logger.error("#### Query type was: " + parsedQuery.getClass());
227 }
228 }
229
230 } else {
231 System.err.println("#### CoreContainer is empty");
232 logger.error("#### CoreContainer is empty");
233 }
234 } else {
235 System.err.println("#### Not an EmbeddedSolrServer. This shouldn't happen." + solr_core.getClass());
236 logger.error("#### Not an EmbeddedSolrServer. This shouldn't happen." + solr_core.getClass());
237 }
238
239
240 return terms;
241 }
242
243 public SharedSoleneQueryResult runQuery(String query_string)
244 {
245 if (query_string == null || query_string.equals(""))
246 {
247 utf8out.println("The query word is not indicated ");
248 utf8out.flush();
249 return null;
250 }
251
252 SolrQueryResult solr_query_result = new SolrQueryResult();
253 solr_query_result.clear();
254
255 if (_facetQueries.size() > 0)
256 {
257 HashMap<String, ArrayList<String>> grouping = new HashMap<String, ArrayList<String>>();
258 for (String currentQuery : _facetQueries)
259 {
260 //Facet queries are stored in JSON, so we have to decode it
261 Gson gson = new Gson();
262 Type type = new TypeToken<List<String>>()
263 {
264 }.getType();
265 List<String> queryElems = gson.fromJson(currentQuery, type);
266
267 //Group each query segment by the index it uses
268 for (String currentQueryElement : queryElems)
269 {
270 String decodedQueryElement = null;
271 try
272 {
273 decodedQueryElement = URLDecoder.decode(currentQueryElement, "UTF-8");
274 }
275 catch (Exception ex)
276 {
277 continue;
278 }
279
280 int colonIndex = currentQueryElement.indexOf(":");
281 String indexShortName = currentQueryElement.substring(0, colonIndex);
282
283 if (grouping.get(indexShortName) == null)
284 {
285 grouping.put(indexShortName, new ArrayList<String>());
286 }
287 grouping.get(indexShortName).add(decodedQueryElement);
288 }
289 }
290
291 //Construct the facet query string to add to the regular query string
292 StringBuilder facetQueryString = new StringBuilder();
293 int keysetCounter = 0;
294 for (String key : grouping.keySet())
295 {
296 StringBuilder currentFacetString = new StringBuilder("(");
297 int groupCounter = 0;
298 for (String queryElem : grouping.get(key))
299 {
300 currentFacetString.append(queryElem);
301
302 groupCounter++;
303 if (groupCounter < grouping.get(key).size())
304 {
305 currentFacetString.append(" OR ");
306 }
307 }
308 currentFacetString.append(")");
309
310 facetQueryString.append(currentFacetString);
311
312 keysetCounter++;
313 if (keysetCounter < grouping.keySet().size())
314 {
315 facetQueryString.append(" AND ");
316 }
317 }
318
319 if (facetQueryString.length() > 0)
320 {
321 query_string += " AND " + facetQueryString;
322 }
323 }
324
325
326 SolrQuery solrQuery = new SolrQuery(query_string);
327 solrQuery.addSort(this.sort_field, SolrQuery.ORDER.valueOf(this.sort_order)); // sort param, like "score desc" or "byORG asc"
328 solrQuery.setStart(start_results); // which result to start from
329 solrQuery.setRows((end_results - start_results) + 1); // how many results per "page"
330
331 // http://lucene.472066.n3.nabble.com/get-term-frequency-just-only-keywords-search-td4084510.html
332 // WORKS (search didx core):
333 //TI:farming
334 //docOID,score,termfreq(TI,'farming'),totaltermfreq(TI,'farming')
335
336
337 // which fields to return for each document, we'll add the request for totaltermfreq later
338 // fl=docOID score termfreq(TI,'farming') totaltermfreq(TI,'farming')
339 solrQuery.setFields("docOID", "score"); //solrParams.set("fl", "docOID score totaltermfreq(field,'queryterm')");
340
341 //solrQuery.setTerms(true); // turn on the termsComponent
342 //solrQuery.set("terms.fl", "ZZ"); // which field to get the terms from. ModifiableSolrParams method
343
344 // http://wiki.apache.org/solr/TermVectorComponent and https://cwiki.apache.org/confluence/display/solr/The+Term+Vector+Component
345 // http://lucene.472066.n3.nabble.com/get-term-frequency-just-only-keywords-search-td4084510.html
346 // http://stackoverflow.com/questions/13031534/word-frequency-in-solr
347 // http://wiki.apache.org/solr/FunctionQuery#tf and #termfreq and #totaltermfreq
348 // https://wiki.apache.org/solr/TermsComponent
349
350 //solrParams.set("tv.tf", true);// turn on the terms vector Component
351 //solrParams.set("tv.fl", "ZZ");// which field to get the terms from /// ZZ
352
353
354 if (_facets.size() > 0)
355 {
356 // enable facet counts in the query response
357 solrQuery.setFacet(true); //solrParams.set("facet", "true");
358 for (int i = 0; i < _facets.size(); i++)
359 {
360 // add this field as a facet
361 solrQuery.addFacetField(_facets.get(i)); // solrParams.add("facet.field", _facets.get(i));
362 }
363 }
364
365 // get the individual terms that make up the query, then request solr to return the totaltermfreq for each term
366 Term[] terms = getTerms(solrQuery, query_string);
367 if(terms != null) {
368 for(int i = 0; i < terms.length; i++) {
369 Term term = terms[i];
370 String field = term.field();
371 String queryTerm = term.text();
372 // totaltermfreq(TI, 'farming') termfreq(TI, 'farming')
373
374 solrQuery.addField("totaltermfreq(" + field + ",'" + queryTerm + "')");
375 solrQuery.addField("termfreq(" + field + ",'" + queryTerm + "')");
376 }
377 }
378
379 // do the query
380 try
381 {
382 QueryResponse solrResponse = solr_core.query(solrQuery); //solr_core.query(solrParams);
383 SolrDocumentList hits = solrResponse.getResults();
384 //TermsResponse termResponse = solrResponse.getTermsResponse(); // null unless termvectors=true in schema.xml
385
386 if (hits != null)
387 {
388 logger.info("*** hits size = " + hits.size());
389 logger.info("*** num docs found = " + hits.getNumFound());
390
391 logger.info("*** start results = " + start_results);
392 logger.info("*** end results = " + end_results);
393 logger.info("*** max docs = " + max_docs);
394
395 // numDocsFound is the total number of matching docs in the collection
396 // as opposed to the number of documents returned in the hits list
397
398 solr_query_result.setTotalDocs((int) hits.getNumFound());
399
400 solr_query_result.setStartResults(start_results);
401 solr_query_result.setEndResults(start_results + hits.size());
402
403
404 // get the first field we're searching in, this will be the fallback field
405 int sepIndex = query_string.indexOf(":");
406 String defaultField = query_string.substring(0, sepIndex);
407 //String query = query_string.substring(sepIndex + 2, query_string.length() - 1); // Replaced by call to getTerms()
408
409 //solr_query_result.addTerm(query, field, (int) hits.getNumFound(), -1);
410
411 // Output the matching documents
412 for (int i = 0; i < hits.size(); i++)
413 {
414 SolrDocument doc = hits.get(i);
415
416 // Need to think about how to support document term frequency. Make zero for now
417 int doc_term_freq = 0;
418 String docOID = (String) doc.get("docOID");
419 Float score = (Float) doc.get("score");
420
421 logger.info("**** docOID = " + docOID);
422 logger.info("**** score = " + score);
423
424
425 // solr returns each term's totaltermfreq, ttf, at the document level, even though
426 // the ttf is the same for each document. So extract this information just for the first document
427 if(i == 0) { // first document
428
429 if(terms != null) {
430 for(int j = 0; j < terms.length; j++) {
431 Term term = terms[j];
432 String field = term.field();
433 String queryTerm = term.text();
434
435 // totaltermfreq(TI, 'farming') termfreq(TI, 'farming')
436 Long totaltermfreq = (Long)doc.get("totaltermfreq("+field+",'"+queryTerm+"')");
437 Integer termfreq = (Integer)doc.get("termfreq("+field+",'"+queryTerm+"')");
438
439 //System.err.println("**** ttf = " + totaltermfreq);
440 //System.err.println("**** tf = " + termfreq);
441 //logger.info("**** ttf = " + totaltermfreq);
442 //logger.info("**** tf = " + termfreq);
443
444 solr_query_result.addTerm(queryTerm, field, (int) hits.getNumFound(), totaltermfreq.intValue()); // long totaltermfreq to int
445 }
446 } else { // no terms extracted from query_string
447 solr_query_result.addTerm(query_string, defaultField, (int) hits.getNumFound(), -1); // no terms
448 }
449 }
450
451 solr_query_result.addDoc(docOID, score.floatValue(), doc_term_freq); // doc_termfreq for which term????
452 }
453 }
454 else
455 {
456 solr_query_result.setTotalDocs(0);
457
458 solr_query_result.setStartResults(0);
459 solr_query_result.setEndResults(0);
460 }
461
462 solr_query_result.setFacetResults(solrResponse.getFacetFields());
463 }
464 catch (SolrServerException server_exception)
465 {
466 server_exception.printStackTrace();
467 solr_query_result.setError(SolrQueryResult.SERVER_ERROR);
468 }
469
470 return solr_query_result;
471 }
472
473 //Greenstone universe operates with a base of 1 for "start_results"
474 //But Solr operates from 0
475 public void setStartResults(int start_results)
476 {
477 if (start_results < 0)
478 {
479 start_results = 0;
480 }
481 this.start_results = start_results - 1;
482 }
483
484 public void cleanUp()
485 {
486 super.cleanUp();
487 }
488
489}
Note: See TracBrowser for help on using the repository browser.