cdmlib-services/src/test/java/eu/etaxonomy/cdm/api/service/TaxonServiceSearchTest.java

   1 /**
   2  * Copyright (C) 2009 EDIT
   3  * European Distributed Institute of Taxonomy
   4  * http://www.e-taxonomy.eu
   5  *
   6  * The contents of this file are subject to the Mozilla Public License Version 1.1
   7  * See LICENSE.TXT at the top of this package for the full license terms.
   8  */
   9
  10 package eu.etaxonomy.cdm.api.service;
  11
  12 import static org.junit.Assert.assertEquals;
  13 import static org.junit.Assert.assertNotNull;
  14
  15 import java.io.FileNotFoundException;
  16 import java.io.IOException;
  17 import java.util.ArrayList;
  18 import java.util.Arrays;
  19 import java.util.EnumSet;
  20 import java.util.HashSet;
  21 import java.util.List;
  22 import java.util.Map;
  23 import java.util.Set;
  24 import java.util.UUID;
  25
  26 import org.apache.commons.lang.RandomStringUtils;
  27 import org.apache.log4j.Level;
  28 import org.apache.log4j.Logger;
  29 import org.apache.lucene.document.Document;
  30 import org.apache.lucene.index.CorruptIndexException;
  31 import org.apache.lucene.queryparser.classic.ParseException;
  32 import org.junit.Assert;
  33 import org.junit.Before;
  34 import org.junit.Ignore;
  35 import org.junit.Test;
  36 import org.unitils.dbunit.annotation.DataSet;
  37 import org.unitils.spring.annotation.SpringBeanByType;
  38
  39 import eu.etaxonomy.cdm.api.service.config.FindTaxaAndNamesConfiguratorImpl;
  40 import eu.etaxonomy.cdm.api.service.config.IFindTaxaAndNamesConfigurator;
  41 import eu.etaxonomy.cdm.api.service.pager.Pager;
  42 import eu.etaxonomy.cdm.api.service.search.ICdmMassIndexer;
  43 import eu.etaxonomy.cdm.api.service.search.LuceneMultiSearchException;
  44 import eu.etaxonomy.cdm.api.service.search.SearchResult;
  45 import eu.etaxonomy.cdm.common.UTF8;
  46 import eu.etaxonomy.cdm.common.monitor.DefaultProgressMonitor;
  47 import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper;
  48 import eu.etaxonomy.cdm.model.common.CdmBase;
  49 import eu.etaxonomy.cdm.model.common.DefinedTermBase;
  50 import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
  51 import eu.etaxonomy.cdm.model.common.Language;
  52 import eu.etaxonomy.cdm.model.description.CategoricalData;
  53 import eu.etaxonomy.cdm.model.description.CommonTaxonName;
  54 import eu.etaxonomy.cdm.model.description.DescriptionBase;
  55 import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
  56 import eu.etaxonomy.cdm.model.description.Distribution;
  57 import eu.etaxonomy.cdm.model.description.Feature;
  58 import eu.etaxonomy.cdm.model.description.PresenceAbsenceTerm;
  59 import eu.etaxonomy.cdm.model.description.State;
  60 import eu.etaxonomy.cdm.model.description.StateData;
  61 import eu.etaxonomy.cdm.model.description.TaxonDescription;
  62 import eu.etaxonomy.cdm.model.description.TextData;
  63 import eu.etaxonomy.cdm.model.location.Country;
  64 import eu.etaxonomy.cdm.model.location.NamedArea;
  65 import eu.etaxonomy.cdm.model.name.BotanicalName;
  66 import eu.etaxonomy.cdm.model.name.NonViralName;
  67 import eu.etaxonomy.cdm.model.name.Rank;
  68 import eu.etaxonomy.cdm.model.name.TaxonNameBase;
  69 import eu.etaxonomy.cdm.model.reference.Reference;
  70 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  71 import eu.etaxonomy.cdm.model.taxon.Classification;
  72 import eu.etaxonomy.cdm.model.taxon.Synonym;
  73 import eu.etaxonomy.cdm.model.taxon.SynonymType;
  74 import eu.etaxonomy.cdm.model.taxon.Taxon;
  75 import eu.etaxonomy.cdm.model.taxon.TaxonBase;
  76 import eu.etaxonomy.cdm.model.taxon.TaxonNode;
  77 import eu.etaxonomy.cdm.model.taxon.TaxonRelationship;
  78 import eu.etaxonomy.cdm.model.taxon.TaxonRelationshipType;
  79 import eu.etaxonomy.cdm.persistence.dto.UuidAndTitleCache;
  80 import eu.etaxonomy.cdm.persistence.query.MatchMode;
  81 import eu.etaxonomy.cdm.persistence.query.OrderHint;
  82 import eu.etaxonomy.cdm.test.integration.CdmTransactionalIntegrationTest;
  83 import eu.etaxonomy.cdm.test.unitils.CleanSweepInsertLoadStrategy;
  84
  85 /**
  86  * @author a.babadshanjan, a.kohlbecker
  87  * @created 04.02.2009
  88  */
  89 public class TaxonServiceSearchTest extends CdmTransactionalIntegrationTest {
  90
  91     private static final String ABIES_BALSAMEA_UUID = "f65d47bd-4f49-4ab1-bc4a-bc4551eaa1a8";
  92
  93     private static final String ABIES_ALBA_UUID = "7dbd5810-a3e5-44b6-b563-25152b8867f4";
  94
  95     private static final String CLASSIFICATION_UUID = "2a5ceebb-4830-4524-b330-78461bf8cb6b";
  96
  97     private static final String CLASSIFICATION_ALT_UUID = "d7c741e3-ae9e-4a7d-a566-9e3a7a0b51ce";
  98
  99     private static final String D_ABIES_BALSAMEA_UUID = "900108d8-e6ce-495e-b32e-7aad3099135e";
 100
 101     private static final String D_ABIES_ALBA_UUID = "ec8bba03-d993-4c85-8472-18b14942464b";
 102
 103     private static final String D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID = "e9d8c2fd-6409-46d5-9c2e-14a2bbb1b2b1";
 104     private static final int NUM_OF_NEW_RADOM_ENTITIES = 1000;
 105
 106     private static Logger logger = Logger.getLogger(TaxonServiceSearchTest.class);
 107
 108
 109
 110     @SpringBeanByType
 111     private ITaxonService taxonService;
 112     @SpringBeanByType
 113     private ITermService termService;
 114     @SpringBeanByType
 115     private IClassificationService classificationService;
 116     @SpringBeanByType
 117     private IReferenceService referenceService;
 118     @SpringBeanByType
 119     private IDescriptionService descriptionService;
 120     @SpringBeanByType
 121     private INameService nameService;
 122     @SpringBeanByType
 123     private ICdmMassIndexer indexer;
 124
 125     @SpringBeanByType
 126     private ITaxonNodeService nodeService;
 127
 128     private static final int BENCHMARK_ROUNDS = 300;
 129
 130     private Set<Class<? extends CdmBase>> typesToIndex = null;
 131
 132     private NamedArea germany;
 133     private NamedArea france ;
 134     private NamedArea russia ;
 135     private NamedArea canada ;
 136
 137     /**
 138      * @throws java.lang.Exception
 139      */
 140     @Before
 141     public void setUp() throws Exception {
 142         typesToIndex = new HashSet<Class<? extends CdmBase>>();
 143         typesToIndex.add(DescriptionElementBase.class);
 144         typesToIndex.add(TaxonBase.class);
 145         typesToIndex.add(TaxonRelationship.class);
 146
 147         germany =  Country.GERMANY();
 148         france = Country.FRANCEFRENCHREPUBLIC();
 149         russia = Country.RUSSIANFEDERATION();
 150         canada = Country.CANADA();
 151
 152
 153     }
 154
 155     @Test
 156     public void testDbUnitUsageTest() throws Exception {
 157         assertNotNull("taxonService should exist", taxonService);
 158         assertNotNull("nameService should exist", nameService);
 159     }
 160
 161     /**
 162      * Test method for
 163      * {@link eu.etaxonomy.cdm.api.service.TaxonServiceImpl#findTaxaAndNames(eu.etaxonomy.cdm.api.service.config.IFindTaxaAndNamesConfigurator)}
 164      * .
 165      */
 166     @Test
 167     @DataSet
 168     public final void testFindTaxaAndNames() {
 169
 170         // pass 1
 171         IFindTaxaAndNamesConfigurator<?> configurator = new FindTaxaAndNamesConfiguratorImpl();
 172         configurator.setTitleSearchString("Abies*");
 173         configurator.setMatchMode(MatchMode.BEGINNING);
 174         configurator.setDoTaxa(true);
 175         configurator.setDoSynonyms(true);
 176         configurator.setDoNamesWithoutTaxa(true);
 177         configurator.setDoTaxaByCommonNames(true);
 178
 179         Pager<IdentifiableEntity> pager = taxonService.findTaxaAndNames(configurator);
 180         List<IdentifiableEntity> list = pager.getRecords();
 181
 182         if (logger.isDebugEnabled()) {
 183             for (int i = 0; i < list.size(); i++) {
 184                 String nameCache = "";
 185                 if (list.get(i) instanceof NonViralName) {
 186                     nameCache = ((NonViralName<?>) list.get(i)).getNameCache();
 187                 } else if (list.get(i) instanceof TaxonBase) {
 188                     TaxonNameBase<?,?> taxonNameBase = ((TaxonBase) list.get(i)).getName();
 189                     nameCache = HibernateProxyHelper.deproxy(taxonNameBase, NonViralName.class).getNameCache();
 190                 } else {
 191                 }
 192                 logger.debug(list.get(i).getClass() + "(" + i + ")" + ": Name Cache = " + nameCache + ", Title Cache = "
 193                         + list.get(i).getTitleCache());
 194             }
 195         }
 196
 197         logger.debug("number of taxa: " + list.size());
 198         assertEquals(10, list.size());
 199         configurator.setTitleSearchString("Balsam-Tanne");
 200         pager = taxonService.findTaxaAndNames(configurator);
 201         list = pager.getRecords();
 202         assertEquals(1, list.size());
 203         // pass 2
 204         configurator.setDoTaxaByCommonNames(false);
 205         configurator.setDoMisappliedNames(true);
 206         configurator.setClassification(classificationService.load(UUID.fromString(CLASSIFICATION_UUID)));
 207         pager = taxonService.findTaxaAndNames(configurator);
 208         list = pager.getRecords();
 209         assertEquals(0, list.size());
 210
 211     }
 212
 213     /**
 214      * Test method for
 215      * {@link eu.etaxonomy.cdm.api.service.TaxonServiceImpl#findTaxaAndNames(eu.etaxonomy.cdm.api.service.config.IFindTaxaAndNamesConfigurator)}
 216      * .
 217      */
 218     @Test
 219     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 220     public final void testFindTaxaAndNamesWithHybridFormula() {
 221
 222         // pass 1
 223         IFindTaxaAndNamesConfigurator<?> configurator = new FindTaxaAndNamesConfiguratorImpl();
 224         configurator.setTitleSearchString("Achillea*");
 225         configurator.setMatchMode(MatchMode.BEGINNING);
 226         configurator.setDoTaxa(true);
 227         configurator.setDoSynonyms(true);
 228         configurator.setDoNamesWithoutTaxa(true);
 229         configurator.setDoTaxaByCommonNames(true);
 230
 231         Pager<IdentifiableEntity> pager = taxonService.findTaxaAndNames(configurator);
 232     //    Assert.assertEquals("Expecting one taxon",1,pager.getRecords().size());
 233         List<IdentifiableEntity> list = pager.getRecords();
 234     }
 235
 236     /**
 237      * Test method for
 238      * {@link eu.etaxonomy.cdm.api.service.TaxonServiceImpl#searchTaxaByName(java.lang.String, eu.etaxonomy.cdm.model.reference.Reference)}
 239      * .
 240      */
 241     @Test
 242     @DataSet
 243     public final void testSearchTaxaByName() {
 244          IFindTaxaAndNamesConfigurator<?> configurator = new FindTaxaAndNamesConfiguratorImpl();
 245          configurator.setTitleSearchString("Abies bor*");
 246          configurator.setMatchMode(MatchMode.BEGINNING);
 247          configurator.setDoTaxa(true);
 248          configurator.setDoSynonyms(false);
 249          configurator.setDoNamesWithoutTaxa(true);
 250          configurator.setDoTaxaByCommonNames(false);
 251
 252         List<UuidAndTitleCache<IdentifiableEntity>> list = taxonService.findTaxaAndNamesForEditor(configurator);
 253
 254          Assert.assertEquals("Expecting one entity", 1, list.size());
 255
 256          configurator.setTitleSearchString("silver fir");
 257          configurator.setMatchMode(MatchMode.BEGINNING);
 258          configurator.setDoTaxa(false);
 259          configurator.setDoSynonyms(false);
 260          configurator.setDoNamesWithoutTaxa(true);
 261          configurator.setDoTaxaByCommonNames(true);
 262
 263          list = taxonService.findTaxaAndNamesForEditor(configurator);
 264
 265          Assert.assertEquals("Expecting one entity", 1, list.size());
 266
 267     }
 268
 269     @SuppressWarnings("rawtypes")
 270     @Test
 271     @DataSet
 272     public final void testPurgeAndReindex() throws CorruptIndexException, IOException, ParseException {
 273
 274         refreshLuceneIndex();
 275
 276         Pager<SearchResult<TaxonBase>> pager;
 277
 278         pager = taxonService.findByFullText(null, "Abies", null, null, true, null, null, null, null); // --> 8
 279         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 280
 281         indexer.purge(null);
 282         commitAndStartNewTransaction(null);
 283
 284         pager = taxonService.findByFullText(null, "Abies", null, null, true, null, null, null, null); // --> 0
 285         Assert.assertEquals("Expecting no entities since the index has been purged", 0, pager.getCount().intValue());
 286
 287         indexer.reindex(indexer.indexedClasses(), null);
 288         commitAndStartNewTransaction(null);
 289
 290         pager = taxonService.findByFullText(null, "Abies", null, null, true, null, null, null, null); // --> 8
 291         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 292     }
 293
 294
 295     @SuppressWarnings("rawtypes")
 296     @Test
 297     @DataSet
 298     public final void testFindByDescriptionElementFullText_CommonName() throws CorruptIndexException, IOException,
 299             ParseException {
 300
 301         refreshLuceneIndex();
 302
 303         Pager<SearchResult<TaxonBase>> pager;
 304
 305         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne", null, null, null,
 306                 false, null, null, null, null);
 307         Assert.assertEquals("Expecting one entity when searching for CommonTaxonName", 1,
 308                 pager.getCount().intValue());
 309
 310         // the description containing the Nulltanne has no taxon attached,
 311         // taxon.id = null
 312         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Nulltanne", null, null, null,
 313                 false, null, null, null, null);
 314         Assert.assertEquals("Expecting no entity when searching for 'Nulltanne' ", 0, pager.getCount().intValue());
 315
 316         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne", null, null,
 317                 Arrays.asList(new Language[] { Language.GERMAN() }), false, null, null, null, null);
 318         Assert.assertEquals("Expecting one entity when searching in German", 1, pager.getCount().intValue());
 319
 320         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne", null, null,
 321                 Arrays.asList(new Language[] { Language.RUSSIAN() }), false, null, null, null, null);
 322         Assert.assertEquals("Expecting no entity when searching in Russian", 0, pager.getCount().intValue());
 323
 324     }
 325
 326     @SuppressWarnings("rawtypes")
 327     @Test
 328     @DataSet
 329     public final void testFindByDescriptionElementFullText_Distribution() throws CorruptIndexException, IOException, ParseException {
 330
 331         refreshLuceneIndex();
 332
 333         Pager<SearchResult<TaxonBase>> pager;
 334         // by Area
 335         pager = taxonService.findByDescriptionElementFullText(null, "Canada", null, null, null, false, null, null, null, null);
 336         Assert.assertEquals("Expecting one entity when searching for arae 'Canada'", 1, pager.getCount().intValue());
 337         // by Status
 338         pager = taxonService.findByDescriptionElementFullText(null, "present", null, null, null, false, null, null, null, null);
 339         Assert.assertEquals("Expecting one entity when searching for status 'present'", 1, pager.getCount().intValue());
 340     }
 341
 342     @SuppressWarnings("rawtypes")
 343     @Test
 344     @DataSet
 345     public final void testFindByDescriptionElementFullText_wildcard() throws CorruptIndexException, IOException, ParseException {
 346
 347         refreshLuceneIndex();
 348
 349         Pager<SearchResult<TaxonBase>> pager;
 350
 351         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"*", null, null, null, false, null, null, null, null);
 352         Assert.assertEquals("Expecting one entity when searching for CommonTaxonName", 1, pager.getCount().intValue());
 353     }
 354
 355     /**
 356      * Regression test for #3113 (hibernate search: wildcard query can cause BooleanQuery$TooManyClauses: maxClauseCount is set to 1024)
 357      *
 358      * @throws CorruptIndexException
 359      * @throws IOException
 360      * @throws ParseException
 361      */
 362     @SuppressWarnings("rawtypes")
 363     @Test
 364     @DataSet
 365     public final void testFindByDescriptionElementFullText_TooManyClauses() throws CorruptIndexException, IOException, ParseException {
 366
 367         // generate 1024 terms to reproduce the bug
 368         TaxonDescription description = (TaxonDescription) descriptionService.find(UUID.fromString(D_ABIES_ALBA_UUID));
 369         Set<String> uniqueRandomStrs = new HashSet<String>(1024);
 370         while(uniqueRandomStrs.size() < 1024){
 371             uniqueRandomStrs.add(RandomStringUtils.random(10, true, false));
 372         }
 373         for(String rndStr: uniqueRandomStrs){
 374             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 375         }
 376         descriptionService.saveOrUpdate(description);
 377         commitAndStartNewTransaction(null);
 378
 379         refreshLuceneIndex();
 380
 381         Pager<SearchResult<TaxonBase>> pager;
 382
 383         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, false, null, null, null, null);
 384         Assert.assertEquals("Expecting all 1024 entities grouped into one SearchResult item when searching for Rot*", 1, pager.getCount().intValue());
 385     }
 386
 387     /**
 388      * Regression test for #3116 (fulltext search: always only one page of results)
 389      *
 390      * @throws CorruptIndexException
 391      * @throws IOException
 392      * @throws ParseException
 393      */
 394     @SuppressWarnings("rawtypes")
 395     @Test
 396     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 397     @Ignore
 398     public final void testFullText_Paging() throws CorruptIndexException, IOException, ParseException {
 399
 400         Reference sec = ReferenceFactory.newDatabase();
 401         referenceService.save(sec);
 402
 403         Set<String> uniqueRandomStrs = new HashSet<String>(1024);
 404         int numOfItems = 100;
 405         while(uniqueRandomStrs.size() < numOfItems){
 406             uniqueRandomStrs.add(RandomStringUtils.random(5, true, false));
 407         }
 408
 409         for(String rndStr: uniqueRandomStrs){
 410
 411             Taxon taxon = Taxon.NewInstance(BotanicalName.NewInstance(Rank.SERIES()), sec);
 412             taxon.setTitleCache("Tax" + rndStr, true);
 413             taxonService.save(taxon);
 414
 415             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 416             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 417             descriptionService.saveOrUpdate(description);
 418         }
 419
 420         commitAndStartNewTransaction(new String[]{"TAXONBASE"});
 421         refreshLuceneIndex();
 422
 423         int pageSize = 10;
 424
 425         Pager<SearchResult<TaxonBase>> pager;
 426
 427         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, false, pageSize, null, null, null);
 428         Assert.assertEquals("unexpeted number of pages", Integer.valueOf(numOfItems / pageSize), pager.getPagesAvailable());
 429         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, false, pageSize, 9, null, null);
 430         Assert.assertNotNull("last page must have records", pager.getRecords());
 431         Assert.assertNotNull("last item on last page must exist", pager.getRecords().get(0));
 432         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, false, pageSize, 10, null, null);
 433         Assert.assertNotNull("last page + 1 must not have any records", pager.getRecords());
 434     }
 435
 436     /**
 437      * test for max score and sort by score of hit groups
 438      * with all matches per taxon in a single TextData  element
 439      * see {@link #testFullText_ScoreAndOrder_2()} for the complement
 440      * test with matches in multiple TextData per taxon
 441      *
 442      * @throws CorruptIndexException
 443      * @throws IOException
 444      * @throws ParseException
 445      */
 446     @SuppressWarnings("rawtypes")
 447     @Test
 448     @DataSet
 449     @Ignore // test fails, maybe the assumptions made here are not compatible with the lucene scoring mechanism see http://lucene.apache.org/core/3_6_1/scoring.html
 450     public final void testFullText_ScoreAndOrder_1() throws CorruptIndexException, IOException, ParseException {
 451
 452         int numOfTaxa = 3;
 453
 454         UUID[] taxonUuids = new UUID[numOfTaxa];
 455         StringBuilder text = new StringBuilder();
 456
 457         for(int i = 0; i < numOfTaxa; i++){
 458
 459             Taxon taxon = Taxon.NewInstance(BotanicalName.NewInstance(null), null);
 460             taxon.setTitleCache("Taxon_" + i, true);
 461             taxonUuids[i] = taxon.getUuid();
 462             taxonService.save(taxon);
 463
 464             text.append(" ").append("Rot");
 465             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 466             description.addElement(TextData.NewInstance(text.toString(), Language.DEFAULT(), null));
 467             descriptionService.saveOrUpdate(description);
 468         }
 469
 470         commitAndStartNewTransaction(null);
 471         refreshLuceneIndex();
 472
 473         Pager<SearchResult<TaxonBase>> pager;
 474
 475         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Rot", null, null, null, false, null, null, null, null);
 476         for(int i = 0; i < numOfTaxa; i++){
 477             Assert.assertEquals("taxa should be orderd by relevance (= score)", taxonUuids[numOfTaxa - i - 1], pager.getRecords().get(i).getEntity().getUuid());
 478         }
 479         Assert.assertEquals("max score should be equal to the score of the first element", pager.getRecords().get(0).getMaxScore(), pager.getRecords().get(0).getScore(), 0);
 480     }
 481
 482     /**
 483      * test for max score and sort by score of hit groups
 484      * with all matches per taxon in a multiple TextData elements
 485      * see {@link #testFullText_ScoreAndOrder_1()} for the complement
 486      * test with matches in a single TextData per taxon
 487      *
 488      * @throws CorruptIndexException
 489      * @throws IOException
 490      * @throws ParseException
 491      */
 492     @SuppressWarnings("rawtypes")
 493     @Test
 494     @DataSet
 495     @Ignore // test fails, maybe the assumptions made here are not compatible with the lucene scoring mechanism see http://lucene.apache.org/core/3_6_1/scoring.html
 496     public final void testFullText_ScoreAndOrder_2() throws CorruptIndexException, IOException, ParseException {
 497
 498         int numOfTaxa = 3;
 499
 500         UUID[] taxonUuids = new UUID[numOfTaxa];
 501
 502         for(int i = 0; i < numOfTaxa; i++){
 503
 504             Taxon taxon = Taxon.NewInstance(BotanicalName.NewInstance(null), null);
 505             taxon.setTitleCache("Taxon_" + i, true);
 506             taxonUuids[i] = taxon.getUuid();
 507             taxonService.save(taxon);
 508
 509             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 510             for(int k = 0; k < i; k++){
 511                 description.addElement(TextData.NewInstance("Rot", Language.DEFAULT(), null));
 512             }
 513             descriptionService.saveOrUpdate(description);
 514         }
 515
 516         commitAndStartNewTransaction(null);
 517         refreshLuceneIndex();
 518
 519         Pager<SearchResult<TaxonBase>> pager;
 520
 521         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Rot", null, null, null, false, null, null, null, null);
 522         for(int i = 0; i < numOfTaxa; i++){
 523             Assert.assertEquals("taxa should be orderd by relevance (= score)", taxonUuids[numOfTaxa - i - 1], pager.getRecords().get(i).getEntity().getUuid());
 524         }
 525         Assert.assertEquals("max score should be equal to the score of the first element", pager.getRecords().get(0).getMaxScore(), pager.getRecords().get(0).getScore(), 0);
 526     }
 527
 528
 529     /**
 530      * @throws CorruptIndexException
 531      * @throws IOException
 532      * @throws ParseException
 533      * @throws LuceneMultiSearchException
 534      */
 535     @Test
 536     @DataSet
 537     public final void testFullText_Grouping() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
 538
 539         TaxonDescription description = (TaxonDescription) descriptionService.find(UUID.fromString(D_ABIES_ALBA_UUID));
 540         Set<String> uniqueRandomStrs = new HashSet<String>(1024);
 541         int numOfItems = 100;
 542         while(uniqueRandomStrs.size() < numOfItems){
 543             uniqueRandomStrs.add(RandomStringUtils.random(5, true, false));
 544         }
 545         for(String rndStr: uniqueRandomStrs){
 546             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 547         }
 548         descriptionService.saveOrUpdate(description);
 549
 550         commitAndStartNewTransaction(new String[]{"DESCRIPTIONELEMENTBASE"});
 551
 552         refreshLuceneIndex();
 553
 554         int pageSize = 10;
 555
 556         Pager<SearchResult<TaxonBase>> pager;
 557         boolean highlightFragments = true;
 558
 559         // test with findByDescriptionElementFullText
 560         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, null, null, highlightFragments, pageSize, null, null, null);
 561         logSearchResults(pager, Level.DEBUG, null);
 562         Assert.assertEquals("All matches should be grouped into a single SearchResult element", 1, pager.getRecords().size());
 563         Assert.assertEquals("The count property of the pager must be set correctly", 1, pager.getCount().intValue());
 564         Map<String, String[]> highlightMap = pager.getRecords().get(0).getFieldHighlightMap();
 565         // maxDocsPerGroup is defined in LuceneSearch and defaults to 10
 566         int maxDocsPerGroup = 10;
 567         Assert.assertEquals("expecting 10 highlighted fragments of field 'name'", maxDocsPerGroup, highlightMap.get("name").length);
 568
 569         // test with findByEverythingFullText
 570         pager = taxonService.findByEverythingFullText( "Rot*", null, null, highlightFragments, pageSize, null, null, null);
 571         logSearchResults(pager, Level.DEBUG, null);
 572         Assert.assertEquals("All matches should be grouped into a single SearchResult element", 1, pager.getRecords().size());
 573         Assert.assertEquals("The count property of the pager must be set correctly", 1, pager.getCount().intValue());
 574         highlightMap = pager.getRecords().get(0).getFieldHighlightMap();
 575         // maxDocsPerGroup is defined in LuceneSearch and defaults to 10
 576         maxDocsPerGroup = 10;
 577         Assert.assertEquals("expecting 10 highlighted fragments of field 'name'", maxDocsPerGroup, highlightMap.get("name").length);
 578
 579     }
 580
 581     @SuppressWarnings("rawtypes")
 582     @Test
 583     @DataSet
 584     @Ignore
 585     public final void testFindByDescriptionElementFullText_TextData() throws CorruptIndexException, IOException, ParseException {
 586
 587         refreshLuceneIndex();
 588
 589         Pager<SearchResult<TaxonBase>> pager;
 590         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Abies", null, null, null, false, null, null, null, null);
 591         logSearchResults(pager, Level.DEBUG, null);
 592         Assert.assertEquals("Expecting one entity when searching for any TextData", 1, pager.getCount().intValue());
 593         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 594         Assert.assertTrue("Expecting two docs, one for RUSSIAN and one for GERMAN", pager.getRecords().get(0).getDocs().size() == 2);
 595         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getDocs().iterator().next().get("inDescription.taxon.titleCache"));
 596
 597
 598         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, null, null, false, null, null, null, null);
 599         Assert.assertEquals("Expecting one entity when searching for any type", 1, pager.getCount().intValue());
 600
 601         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, Arrays.asList(new Feature[]{Feature.UNKNOWN()}), null, false, null, null, null, null);
 602         Assert.assertEquals("Expecting one entity when searching for any type and for Feature DESCRIPTION", 1, pager.getCount().intValue());
 603
 604         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, Arrays.asList(new Feature[]{Feature.CHROMOSOME_NUMBER()}), null, false, null, null, null, null);
 605         Assert.assertEquals("Expecting no entity when searching for any type and for Feature CHROMOSOME_NUMBER", 0, pager.getCount().intValue());
 606
 607         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, Arrays.asList(new Feature[]{Feature.CHROMOSOME_NUMBER(), Feature.UNKNOWN()}), null, false, null, null, null, null);
 608         Assert.assertEquals("Expecting no entity when searching for any type and for Feature DESCRIPTION or CHROMOSOME_NUMBER", 1, pager.getCount().intValue());
 609
 610         pager = taxonService.findByDescriptionElementFullText(Distribution.class, "Abies", null, null, null, false, null, null, null, null);
 611         Assert.assertEquals("Expecting no entity when searching for Distribution", 0, pager.getCount().intValue());
 612
 613         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, null, Arrays.asList(new Language[]{}), false, null, null, null, null);
 614         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 615         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 616
 617         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, null, Arrays.asList(new Language[]{Language.RUSSIAN()}), false, null, null, null, null);
 618         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 619         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 620
 621         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 622         Assert.assertEquals("Expecting no entity", 0, pager.getCount().intValue());
 623
 624         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 625         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 626         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 627     }
 628
 629     @SuppressWarnings("rawtypes")
 630     @Test
 631     @DataSet
 632     public final void testFindByDescriptionElementFullText_MultipleWords() throws CorruptIndexException, IOException, ParseException {
 633
 634         refreshLuceneIndex();
 635
 636         // Pflanzenart aus der Gattung der Tannen
 637         long start = System.currentTimeMillis();
 638
 639         Pager<SearchResult<TaxonBase>> pager;
 640         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Tannen", null, null, null, false, null, null, null, null);
 641         Assert.assertEquals("OR search : Expecting one entity", 1, pager.getCount().intValue());
 642
 643         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Wespen", null, null, null, false, null, null, null, null);
 644         Assert.assertEquals("OR search : Expecting one entity", 1, pager.getCount().intValue());
 645
 646         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Tannen", null, null, null, false, null, null, null, null);
 647         Assert.assertEquals("AND search : Expecting one entity", 1, pager.getCount().intValue());
 648
 649         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Wespen", null, null, null, false, null, null, null, null);
 650         Assert.assertEquals("AND search : Expecting no entity", 0, pager.getCount().intValue());
 651
 652         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Tannen\"", null, null, null, false, null, null, null, null);
 653         Assert.assertEquals("Phrase search : Expecting one entity", 1, pager.getCount().intValue());
 654
 655         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Wespen\"", null, null, null, false, null, null, null, null);
 656         Assert.assertEquals("Phrase search : Expecting one entity", 0, pager.getCount().intValue());
 657
 658         logger.info("testFindByDescriptionElementFullText_MultipleWords() duration: " + (System.currentTimeMillis() - start) + "ms");
 659
 660     }
 661
 662
 663     @SuppressWarnings("rawtypes")
 664     @Test
 665     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 666     public final void testFindByDescriptionElementFullText_modify_DescriptionElement() throws CorruptIndexException, IOException, ParseException {
 667
 668         refreshLuceneIndex();
 669
 670         Pager<SearchResult<TaxonBase>> pager;
 671         //
 672         // modify the DescriptionElement
 673         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 674         Assert.assertTrue("Search did not return any results", pager.getRecords().size() > 0);
 675         Assert.assertTrue("Expecting only one doc", pager.getRecords().get(0).getDocs().size() == 1);
 676         Document indexDocument = pager.getRecords().get(0).getDocs().iterator().next();
 677         String[] descriptionElementUuidStr = indexDocument.getValues("uuid");
 678         String[] inDescriptionUuidStr = indexDocument.getValues("inDescription.uuid");
 679         // is only one uuid!
 680         DescriptionElementBase textData = descriptionService.getDescriptionElementByUuid(UUID.fromString(descriptionElementUuidStr[0]));
 681
 682         ((TextData)textData).removeText(Language.GERMAN());
 683         ((TextData)textData).putText(Language.SPANISH_CASTILIAN(), "abeto bals"+UTF8.SMALL_A_ACUTE+"mico");
 684
 685         descriptionService.saveDescriptionElement(textData);
 686         commitAndStartNewTransaction(null);
 687 //        printDataSet(System.out, new String[] {
 688 //                "DESCRIPTIONELEMENTBASE", "LANGUAGESTRING", "DESCRIPTIONELEMENTBASE_LANGUAGESTRING" }
 689 //        );
 690
 691         //
 692         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 693         Assert.assertEquals("The german 'Balsam-Tanne' TextData should no longer be indexed", 0, pager.getCount().intValue());
 694         pager = taxonService.findByDescriptionElementFullText(TextData.class, "abeto", null, null, Arrays.asList(new Language[]{Language.SPANISH_CASTILIAN()}), false, null, null, null, null);
 695         Assert.assertEquals("expecting to find the SPANISH_CASTILIAN 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico'", 1, pager.getCount().intValue());
 696         pager = taxonService.findByDescriptionElementFullText(TextData.class, "bals"+UTF8.SMALL_A_ACUTE+"mico", null, null, null, false, null, null, null, null);
 697         Assert.assertEquals("expecting to find the SPANISH_CASTILIAN 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico'", 1, pager.getCount().intValue());
 698
 699         //
 700         // modify the DescriptionElement via the Description object
 701         DescriptionBase<?> description = descriptionService.find(UUID.fromString(inDescriptionUuidStr[0]));
 702         Set<DescriptionElementBase> elements = description.getElements();
 703         for( DescriptionElementBase elm : elements){
 704             if(elm.getUuid().toString().equals(descriptionElementUuidStr[0])){
 705                 ((TextData)elm).removeText(Language.SPANISH_CASTILIAN());
 706                 ((TextData)elm).putText(Language.POLISH(), "Jod"+UTF8.POLISH_L+"a balsamiczna");
 707             }
 708         }
 709         descriptionService.saveOrUpdate(description);
 710         commitAndStartNewTransaction(null);
 711         pager = taxonService.findByDescriptionElementFullText(TextData.class, "abeto", null, null, Arrays.asList(new Language[]{Language.SPANISH_CASTILIAN()}), false, null, null, null, null);
 712         Assert.assertEquals("The spanish 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico' TextData should no longer be indexed", 0, pager.getCount().intValue());
 713         pager = taxonService.findByDescriptionElementFullText(TextData.class, "balsamiczna", null, null, Arrays.asList(new Language[]{Language.POLISH()}), false, null, null, null, null);
 714         Assert.assertEquals("expecting to find the POLISH 'Jod"+UTF8.POLISH_L+"a balsamiczna'", 1, pager.getCount().intValue());
 715     }
 716
 717     @SuppressWarnings("rawtypes")
 718     @Test
 719     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 720     public final void testFindByDescriptionElementFullText_modify_Taxon() throws CorruptIndexException, IOException, ParseException {
 721
 722         refreshLuceneIndex();
 723
 724         Pager<SearchResult<TaxonBase>> pager;
 725         Taxon t_abies_balsamea = (Taxon)taxonService.find(UUID.fromString(ABIES_BALSAMEA_UUID));
 726         TaxonDescription d_abies_balsamea = (TaxonDescription)descriptionService.find(UUID.fromString(D_ABIES_BALSAMEA_UUID));
 727
 728         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 729         Assert.assertEquals("expecting to find the GERMAN 'Balsam-Tanne'", 1, pager.getCount().intValue());
 730
 731         // exchange the Taxon with another one via the Taxon object
 732         // 1.) remove existing description:
 733         t_abies_balsamea.removeDescription(d_abies_balsamea);
 734
 735         taxonService.saveOrUpdate(t_abies_balsamea);
 736         commitAndStartNewTransaction(null);
 737
 738         t_abies_balsamea = (Taxon)taxonService.find(t_abies_balsamea.getUuid());
 739
 740         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 741         Assert.assertEquals("'Balsam-Tanne' should no longer be found", 0, pager.getCount().intValue());
 742
 743         // 2.) create new description and add to taxon:
 744         TaxonDescription d_abies_balsamea_new = TaxonDescription.NewInstance();
 745         d_abies_balsamea_new
 746                 .addElement(TextData
 747                         .NewInstance(
 748                                 "Die Balsamtanne ist mit bis zu 30 m Höhe ein mittelgro"+UTF8.SHARP_S+"er Baum und kann bis zu 200 Jahre alt werden",
 749                                 Language.GERMAN(), null));
 750         t_abies_balsamea.addDescription(d_abies_balsamea_new);
 751         // set authorshipCache to null to avoid validation exception,
 752         // this is maybe not needed in future,  see ticket #3344
 753         BotanicalName abies_balsamea = HibernateProxyHelper.deproxy(t_abies_balsamea.getName(), BotanicalName.class);
 754         abies_balsamea.setAuthorshipCache(null);
 755         printDataSet(System.err, new String[] {"LANGUAGESTRING_AUD"});
 756         taxonService.saveOrUpdate(t_abies_balsamea);
 757         commitAndStartNewTransaction(null);
 758
 759 //        printDataSet(System.out, new String[] {
 760 //                "DESCRIPTIONBASE"
 761 //        });
 762
 763         pager = taxonService.findByDescriptionElementFullText(TextData.class, "mittelgro"+UTF8.SHARP_S+"er Baum", null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 764         Assert.assertEquals("the taxon should be found via the new Description", 1, pager.getCount().intValue());
 765     }
 766
 767     @SuppressWarnings("rawtypes")
 768     @Test
 769     @DataSet
 770     public final void testFindByDescriptionElementFullText_modify_Classification() throws CorruptIndexException, IOException, ParseException {
 771
 772         refreshLuceneIndex();
 773
 774         Pager<SearchResult<TaxonBase>> pager;
 775
 776         // put taxon into other classification, new taxon node
 777         Classification classification = classificationService.find(UUID.fromString(CLASSIFICATION_UUID));
 778         Classification alternateClassification = classificationService.find(UUID.fromString(CLASSIFICATION_ALT_UUID));
 779
 780         // TODO: why is the test failing when the childNode is already retrieved here, and not after the following four lines?
 781         //TaxonNode childNode = classification.getChildNodes().iterator().next();
 782
 783         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 784         Assert.assertEquals("expecting to find the GERMAN 'Balsam-Tanne' even if filtering by classification", 1, pager.getCount().intValue());
 785         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", alternateClassification, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 786         Assert.assertEquals("GERMAN 'Balsam-Tanne' should NOT be found in other classification", 0, pager.getCount().intValue());
 787
 788         // check for the right taxon node
 789         TaxonNode childNode = classification.getChildNodes().iterator().next();
 790         Assert.assertEquals("expecting Abies balsamea sec.", childNode.getTaxon().getUuid().toString(), ABIES_BALSAMEA_UUID);
 791         Assert.assertEquals("expecting default classification", childNode.getClassification().getUuid().toString(), CLASSIFICATION_UUID);
 792
 793         // moving the taxon around, the rootnode is only a proxy
 794         alternateClassification.setRootNode(HibernateProxyHelper.deproxy(alternateClassification.getRootNode(), TaxonNode.class));
 795         alternateClassification.addChildNode(childNode, null, null);
 796
 797         classificationService.saveOrUpdate(alternateClassification);
 798         commitAndStartNewTransaction(null);
 799
 800 //        printDataSet(System.out, new String[] {
 801 //            "TAXONBASE", "TAXONNODE", "CLASSIFICATION"
 802 //        });
 803
 804         // reload classification
 805         classification = classificationService.find(UUID.fromString(CLASSIFICATION_UUID));
 806         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", alternateClassification, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 807         Assert.assertEquals("GERMAN 'Balsam-Tanne' should now be found in other classification", 1, pager.getCount().intValue());
 808
 809         classification.getChildNodes().clear();
 810         classificationService.saveOrUpdate(classification);
 811         commitAndStartNewTransaction(null);
 812
 813         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", classification, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 814         Assert.assertEquals("Now the GERMAN 'Balsam-Tanne' should NOT be found in original classification", 0, pager.getCount().intValue());
 815
 816     }
 817
 818     @SuppressWarnings("rawtypes")
 819     @Test
 820     @DataSet
 821     public final void testFindByDescriptionElementFullText_CategoricalData() throws CorruptIndexException, IOException, ParseException {
 822
 823         // add CategoricalData
 824         DescriptionBase d_abies_balsamea = descriptionService.find(UUID.fromString(D_ABIES_BALSAMEA_UUID));
 825         // Categorical data
 826         CategoricalData cdata = CategoricalData.NewInstance();
 827         cdata.setFeature(Feature.DESCRIPTION());
 828         State state = State.NewInstance("green", "green", "gn");
 829
 830         StateData statedata = StateData.NewInstance(state);
 831         statedata.putModifyingText(Language.ENGLISH(), "always, even during winter");
 832         cdata.addStateData(statedata);
 833         d_abies_balsamea.addElement(cdata);
 834
 835         UUID termUUID = termService.save(state).getUuid();
 836         descriptionService.save(d_abies_balsamea);
 837
 838         commitAndStartNewTransaction(null);
 839
 840 //        printDataSet(System.out, new String[] {
 841 //                 "STATEDATA", "STATEDATA_DEFINEDTERMBASE", "STATEDATA_LANGUAGESTRING", "LANGUAGESTRING"});
 842
 843         refreshLuceneIndex();
 844
 845         Pager<SearchResult<TaxonBase>> pager;
 846         pager = taxonService.findByDescriptionElementFullText(CategoricalData.class, "green", null, null, null, false, null, null, null, null);
 847         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 848         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 849         Assert.assertTrue("Expecting only one doc", pager.getRecords().get(0).getDocs().size() == 1);
 850         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getDocs().iterator().next().get("inDescription.taxon.titleCache"));
 851
 852
 853         //TODO modify the StateData
 854         TaxonBase taxon = pager.getRecords().get(0).getEntity();
 855
 856         String newName = "Quercus robur";
 857         taxon.setTitleCache(newName + " sec. ", true);
 858
 859         taxonService.saveOrUpdate(taxon);
 860         commitAndStartNewTransaction(null);
 861
 862         taxon = taxonService.find(taxon.getUuid());
 863         Assert.assertEquals(newName + " sec. ", taxon.getTitleCache());
 864         DefinedTermBase term = termService.find(termUUID);
 865
 866         termService.delete(term);
 867
 868     }
 869
 870     @SuppressWarnings("rawtypes")
 871     @Test
 872     @DataSet
 873     public final void testFindByDescriptionElementFullText_Highlighting() throws CorruptIndexException, IOException, ParseException {
 874
 875         refreshLuceneIndex();
 876
 877         Pager<SearchResult<TaxonBase>> pager;
 878         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Abies", null, null, null, true, null, null, null, null);
 879         Assert.assertEquals("Expecting one entity when searching for any TextData", 1, pager.getCount().intValue());
 880         SearchResult<TaxonBase> searchResult = pager.getRecords().get(0);
 881         Assert.assertTrue("the map of highlighted fragments should contain at least one item", searchResult.getFieldHighlightMap().size() > 0);
 882         String[] fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 883         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Abies</B>"));
 884
 885         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Tannen", null, null, null, true, null, null, null, null);
 886         searchResult = pager.getRecords().get(0);
 887         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 888         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 889         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Pflanzenart</B>") || fragments[0].contains("<B>Tannen</B>"));
 890
 891         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Tannen", null, null, null, true, null, null, null, null);
 892         searchResult = pager.getRecords().get(0);
 893         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 894         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 895         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Pflanzenart</B>") && fragments[0].contains("<B>Tannen</B>"));
 896
 897         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Tannen\"", null, null, null, true, null, null, null, null);
 898         searchResult = pager.getRecords().get(0);
 899         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 900         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 901         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Pflanzenart</B> <B>aus</B> <B>der</B> <B>Gattung</B> <B>der</B> <B>Tannen</B>"));
 902
 903         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Gatt*", null, null, null, true, null, null, null, null);
 904         searchResult = pager.getRecords().get(0);
 905         Assert.assertTrue("Wildcard search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 906         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 907         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Gatt"));
 908     }
 909
 910
 911     @Test
 912     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 913     public final void testFindByFullText() throws CorruptIndexException, IOException, ParseException {
 914
 915         refreshLuceneIndex();
 916
 917         Classification europeanAbiesClassification = classificationService.find(UUID.fromString(CLASSIFICATION_UUID));
 918
 919         Pager<SearchResult<TaxonBase>> pager;
 920
 921         pager = taxonService.findByFullText(null, "Abies", null, null, true, null, null, null, null); // --> 7
 922         logSearchResults(pager, Level.DEBUG, null);
 923         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 924
 925         pager = taxonService.findByFullText(Taxon.class, "Abies", null, null, true, null, null, null, null); // --> 6
 926         Assert.assertEquals("Expecting 7 entities", 7, pager.getCount().intValue());
 927
 928         pager = taxonService.findByFullText(Synonym.class, "Abies", null, null, true, null, null, null, null); // --> 1
 929         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 930
 931         pager = taxonService.findByFullText(TaxonBase.class, "sec", null, null, true, null, null, null, null); // --> 7
 932         Assert.assertEquals("Expecting 8 entities", 9, pager.getCount().intValue());
 933
 934         pager = taxonService.findByFullText(null, "genus", null, null, true, null, null, null, null); // --> 1
 935         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 936
 937         pager = taxonService.findByFullText(Taxon.class, "subalpina", null, null, true, null, null, null, null); // --> 0
 938         Assert.assertEquals("Expecting 0 entities", 0, pager.getCount().intValue());
 939
 940         // synonym in classification ???
 941     }
 942
 943     @Test
 944     @DataSet
 945     public final void testPrepareByAreaSearch() throws IOException, ParseException {
 946
 947         List<PresenceAbsenceTerm> statusFilter = new ArrayList<PresenceAbsenceTerm>();
 948         List<NamedArea> areaFilter = new ArrayList<NamedArea>();
 949         areaFilter.add(germany);
 950         areaFilter.add(canada);
 951         areaFilter.add(russia);
 952
 953         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDistribution(areaFilter, statusFilter, null, 20, 0, null, null);
 954         Assert.assertEquals("Expecting 2 entities", Integer.valueOf(2), Integer.valueOf(pager.getRecords().size()));
 955
 956     }
 957
 958     @Test
 959     @DataSet
 960     public final void testFindTaxaAndNamesByFullText() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
 961
 962         refreshLuceneIndex();
 963
 964         Pager<SearchResult<TaxonBase>> pager;
 965
 966         Classification alternateClassification = classificationService.find(UUID.fromString(CLASSIFICATION_ALT_UUID));
 967
 968
 969         pager = taxonService.findTaxaAndNamesByFullText(
 970                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
 971                 "Abies", null, null, null, null, true, null, null, null, null);
 972 //        logPagerRecords(pager, Level.DEBUG);
 973         Assert.assertEquals("doTaxa & doSynonyms", 8, pager.getCount().intValue());
 974
 975         pager = taxonService.findTaxaAndNamesByFullText(
 976                 EnumSet.allOf(TaxaAndNamesSearchMode.class),
 977                 "Abies", null, null, null, null, true, null, null, null, null);
 978 //        logPagerRecords(pager, Level.DEBUG);
 979         Assert.assertEquals("all search modes", 8, pager.getCount().intValue());
 980
 981         pager = taxonService.findTaxaAndNamesByFullText(
 982                 EnumSet.allOf(TaxaAndNamesSearchMode.class),
 983                 "Abies", alternateClassification, null, null, null, true, null, null, null, null);
 984 //        logPagerRecords(pager, Level.DEBUG);
 985         Assert.assertEquals("all search modes, filtered by alternateClassification", 1, pager.getCount().intValue());
 986
 987         pager = taxonService.findTaxaAndNamesByFullText(
 988                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms),
 989                 "Abies", null, null, null, null, true, null, null, null, null);
 990         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 991         SearchResult<TaxonBase> searchResult = pager.getRecords().get(0);
 992         Assert.assertEquals(Synonym.class, searchResult.getEntity().getClass());
 993         // Abies subalpina sec. Kohlbecker, A., Testcase standart views, 2013
 994
 995
 996         pager = taxonService.findTaxaAndNamesByFullText(
 997                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames),
 998                 "Abies", null, null, null, null, true, null, null, null, null);
 999         Assert.assertEquals("Expecting 0 entity", 0, pager.getCount().intValue());
1000
1001
1002         pager = taxonService.findTaxaAndNamesByFullText(
1003                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames),
1004                 "Tanne", null, null, null, null, true, null, null, null, null);
1005         Assert.assertEquals("Expecting 1 entity", 1, pager.getRecords().size());
1006         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
1007
1008         pager = taxonService.findTaxaAndNamesByFullText(
1009                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1010                 "kawakamii", null, null, null, null, true, null, null, null, null);
1011         logSearchResults(pager, Level.DEBUG, null);
1012         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
1013
1014     }
1015
1016     @Test
1017     @DataSet
1018     public final void testFindTaxaAndNamesByFullText_Sort() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
1019
1020         refreshLuceneIndex();
1021
1022         Pager<SearchResult<TaxonBase>> pager;
1023
1024         List<OrderHint> orderHints = new ArrayList<OrderHint>();
1025
1026         String[] docFields2log = new String[]{"id"};
1027
1028         // SortById
1029         orderHints.addAll(OrderHint.ORDER_BY_ID.asList());
1030         pager = taxonService.findTaxaAndNamesByFullText(
1031                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa),
1032                 "Abies", null, null, null, null, true, null, null, orderHints, null);
1033 //        logSearchResults(pager, Level.DEBUG, docFields2log);
1034         int lastId = -1;
1035         for(SearchResult<TaxonBase> rs : pager.getRecords()){
1036             if(lastId != -1){
1037                 Assert.assertTrue("results not sorted by id", lastId < rs.getEntity().getId());
1038             }
1039             lastId = rs.getEntity().getId();
1040         }
1041
1042         orderHints.addAll(OrderHint.ORDER_BY_ID.asList());
1043         pager = taxonService.findTaxaAndNamesByFullText(
1044                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1045                 "Abies", null, null, null, null, true, null, null, orderHints, null);
1046 //        logSearchResults(pager, Level.DEBUG, docFields2log);
1047
1048         lastId = -1;
1049         for(SearchResult<TaxonBase> rs : pager.getRecords()){
1050             if(lastId != -1){
1051                 Assert.assertTrue("results not sorted by id", lastId < rs.getEntity().getId());
1052             }
1053             lastId = rs.getEntity().getId();
1054         }
1055
1056         // Sortby NOMENCLATURAL_SORT_ORDER TODO make assertions !!!
1057         orderHints.clear();
1058         orderHints.addAll(OrderHint.NOMENCLATURAL_SORT_ORDER.asList());
1059         pager = taxonService.findTaxaAndNamesByFullText(
1060                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1061                 "Abies", null, null, null, null, true, null, null, orderHints, null);
1062         logSearchResults(pager, Level.DEBUG, null);
1063
1064     }
1065
1066     @Test
1067     @DataSet
1068     public final void testFindTaxaAndNamesByFullText_AreaFilter() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
1069
1070         refreshLuceneIndex();
1071
1072         Pager<SearchResult<TaxonBase>> pager;
1073
1074         Set<NamedArea> a_germany_canada_russia = new HashSet<NamedArea>();
1075         a_germany_canada_russia.add(germany);
1076         a_germany_canada_russia.add(canada);
1077         a_germany_canada_russia.add(russia);
1078
1079         Set<NamedArea> a_russia = new HashSet<NamedArea>();
1080         a_russia.add(russia);
1081
1082         Set<PresenceAbsenceTerm> present = new HashSet<PresenceAbsenceTerm>();
1083         present.add(PresenceAbsenceTerm.PRESENT());
1084
1085         Set<PresenceAbsenceTerm> present_native = new HashSet<PresenceAbsenceTerm>();
1086         present_native.add(PresenceAbsenceTerm.PRESENT());
1087         present_native.add(PresenceAbsenceTerm.NATIVE());
1088
1089         Set<PresenceAbsenceTerm> absent = new HashSet<PresenceAbsenceTerm>();
1090         absent.add(PresenceAbsenceTerm.ABSENT());
1091
1092         pager = taxonService.findTaxaAndNamesByFullText(
1093                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa),
1094                 "Abies", null, a_germany_canada_russia, null, null, true, null, null, null, null);
1095         logSearchResults(pager, Level.DEBUG, null);
1096
1097         // abies_kawakamii_sensu_komarov as missapplied name for t_abies_balsamea
1098         pager = taxonService.findTaxaAndNamesByFullText(
1099                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms),
1100                 "Abies", null, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1101         Assert.assertEquals("synonyms with matching area filter", 1, pager.getCount().intValue());
1102
1103         pager = taxonService.findTaxaAndNamesByFullText(
1104                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1105                 "Abies", null, a_germany_canada_russia, null, null, true, null, null, null, null);
1106         logSearchResults(pager, Level.DEBUG, null);
1107         Assert.assertEquals("taxa and synonyms with matching area filter", 3, pager.getCount().intValue());
1108
1109         pager = taxonService.findTaxaAndNamesByFullText(
1110                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1111                 "Abies", null, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1112         Assert.assertEquals("taxa and synonyms with matching area & status filter 1", 3, pager.getCount().intValue());
1113
1114         pager = taxonService.findTaxaAndNamesByFullText(
1115                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1116                 "Abies", null, a_germany_canada_russia, present, null, true, null, null, null, null);
1117         Assert.assertEquals("taxa and synonyms with matching area & status filter 2", 2, pager.getCount().intValue());
1118
1119         pager = taxonService.findTaxaAndNamesByFullText(
1120                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1121                 "Abies", null, a_russia, present, null, true, null, null, null, null);
1122         Assert.assertEquals("taxa and synonyms with non matching area & status filter", 0, pager.getCount().intValue());
1123
1124         pager = taxonService.findTaxaAndNamesByFullText(
1125                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames),
1126                 "Tanne", null, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1127         Assert.assertEquals("ByCommonNames with area filter", 1, pager.getCount().intValue());
1128
1129         // abies_kawakamii_sensu_komarov as misapplied name for t_abies_balsamea
1130         pager = taxonService.findTaxaAndNamesByFullText(
1131                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1132                 "Abies", null, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1133         Assert.assertEquals("misappliedNames with matching area & status filter", 1, pager.getCount().intValue());
1134
1135
1136         // 1. remove existing taxon relation
1137         Taxon t_abies_balsamea = (Taxon)taxonService.find(UUID.fromString(ABIES_BALSAMEA_UUID));
1138         Set<TaxonRelationship> relsTo = t_abies_balsamea.getRelationsToThisTaxon();
1139         Assert.assertEquals(1, relsTo.size());
1140         TaxonRelationship taxonRelation = relsTo.iterator().next();
1141         t_abies_balsamea.removeTaxonRelation(taxonRelation);
1142         taxonService.saveOrUpdate(t_abies_balsamea);
1143         commitAndStartNewTransaction(null);
1144
1145         pager = taxonService.findTaxaAndNamesByFullText(
1146                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1147                 "Abies", null, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1148         Assert.assertEquals("misappliedNames with matching area & status filter, should match nothing now", 0, pager.getCount().intValue());
1149
1150         // 2. now add abies_kawakamii_sensu_komarov as misapplied name for t_abies_alba and search for misapplications in russia: ABSENT
1151         Taxon t_abies_kawakamii_sensu_komarov = (Taxon)taxonService.find(UUID.fromString(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1152         Taxon t_abies_alba = (Taxon)taxonService.find(UUID.fromString(ABIES_ALBA_UUID));
1153         t_abies_alba.addMisappliedName(t_abies_kawakamii_sensu_komarov, null, null);
1154
1155         taxonService.update(t_abies_kawakamii_sensu_komarov);
1156
1157         commitAndStartNewTransaction(null);
1158
1159         pager = taxonService.findTaxaAndNamesByFullText(
1160                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1161                 "Abies", null, a_germany_canada_russia, absent, null, true, null, null, null, null);
1162         Assert.assertEquals("misappliedNames with matching area & status filter, should find one", 1, pager.getCount().intValue());
1163
1164     }
1165
1166     @Test
1167     @DataSet
1168     @Ignore // remove once http://dev.e-taxonomy.eu/trac/ticket/5477 is solved
1169     public final void testFindTaxaAndNamesByFullText_AreaFilter_issue5477() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
1170
1171         Set<NamedArea> a_germany_canada_russia = new HashSet<NamedArea>();
1172         a_germany_canada_russia.add(germany);
1173         a_germany_canada_russia.add(canada);
1174         a_germany_canada_russia.add(russia);
1175
1176
1177         Set<PresenceAbsenceTerm> absent = new HashSet<PresenceAbsenceTerm>();
1178         absent.add(PresenceAbsenceTerm.ABSENT());
1179
1180         Taxon t_abies_kawakamii_sensu_komarov = (Taxon)taxonService.find(UUID.fromString(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1181         Taxon t_abies_alba = (Taxon)taxonService.find(UUID.fromString(ABIES_ALBA_UUID));
1182         t_abies_alba.addMisappliedName(t_abies_kawakamii_sensu_komarov, null, null);
1183
1184         /* Since the upgrade from hibernate search 4 to 5.5
1185          * triggering an update of t_abies_alba is no longer sufficient to also update the
1186          * document of t_abies_kawakamii_sensu_komarov in the lucene index.
1187          * the last test in testFindTaxaAndNamesByFullText_AreaFilter() failed in this case.
1188          * This situation is reproduced here:
1189          */
1190         taxonService.update(t_abies_alba);
1191
1192           commitAndStartNewTransaction(null);
1193
1194           Pager pager = taxonService.findTaxaAndNamesByFullText(
1195                   EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1196                   "Abies", null, a_germany_canada_russia, absent, null, true, null, null, null, null);
1197           Assert.assertEquals("misappliedNames with matching area & status filter, should find one", 1, pager.getCount().intValue());
1198     }
1199
1200
1201     /**
1202      * Regression test for #3119: fulltext search: Entity always null whatever search
1203      *
1204      * @throws CorruptIndexException
1205      * @throws IOException
1206      * @throws ParseException
1207      * @throws LuceneMultiSearchException
1208      */
1209     @Test
1210     @DataSet
1211     public final void testFindByEverythingFullText() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
1212
1213         refreshLuceneIndex();
1214
1215         Pager<SearchResult<TaxonBase>> pager;
1216
1217         // via Taxon
1218         pager = taxonService.findByEverythingFullText("Abies", null, null, true, null, null, null, null);
1219         logSearchResults(pager, Level.DEBUG, null);
1220         Assert.assertTrue("Expecting at least 7 entities for 'Abies'", pager.getCount() > 7);
1221         Assert.assertNotNull("Expecting entity", pager.getRecords().get(0).getEntity());
1222         Assert.assertEquals("Expecting Taxon entity", Taxon.class, pager.getRecords().get(0).getEntity().getClass());
1223
1224         // via DescriptionElement
1225         pager = taxonService.findByEverythingFullText("present", null, null, true, null, null, null, null);
1226         Assert.assertEquals("Expecting one entity when searching for area 'present'", 1, pager.getCount().intValue());
1227         Assert.assertNotNull("Expecting entity", pager.getRecords().get(0).getEntity());
1228         Assert.assertEquals("Expecting Taxon entity", Taxon.class, CdmBase.deproxy(pager.getRecords().get(0).getEntity()).getClass());
1229         Assert.assertEquals("Expecting Taxon ", ABIES_BALSAMEA_UUID, pager.getRecords().get(0).getEntity().getUuid().toString());
1230
1231     }
1232
1233
1234     @Test
1235     @DataSet
1236     public final void findByEveryThingFullText() throws CorruptIndexException, IOException, ParseException, LuceneMultiSearchException {
1237
1238         refreshLuceneIndex();
1239
1240         Pager<SearchResult<TaxonBase>> pager;
1241
1242         pager = taxonService.findByEverythingFullText("genus", null, null,  false, null, null, null, null); // --> 1
1243         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
1244
1245         //FIXME FAILS: abies balamea is returned twice, see also testFullText_Grouping()
1246         pager = taxonService.findByEverythingFullText("Balsam", null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
1247         logSearchResults(pager, Level.DEBUG, null);
1248         Assert.assertEquals("expecting to find the Abies balsamea via the GERMAN DescriptionElements", 1, pager.getCount().intValue());
1249
1250         pager = taxonService.findByEverythingFullText("Abies", null, null, true, null, null, null, null);
1251         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
1252         SearchResult<TaxonBase> searchResult = pager.getRecords().get(0);
1253         Assert.assertTrue("the map of highlighted fragments should contain at least one item", searchResult.getFieldHighlightMap().size() > 0);
1254         String[] fragments = searchResult.getFieldHighlightMap().values().iterator().next();
1255         Assert.assertTrue("first fragments should contains serch term", fragments[0].toLowerCase().contains("<b>abies</b>"));
1256     }
1257
1258 //    @SuppressWarnings("rawtypes")
1259 //    @Test
1260 //    @DataSet
1261 //    public final void benchmarkFindTaxaAndNamesHql() throws CorruptIndexException, IOException, ParseException {
1262 //
1263 //        createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1264 //
1265 //        IFindTaxaAndNamesConfigurator configurator = new FindTaxaAndNamesConfiguratorImpl();
1266 //        configurator.setTitleSearchString("Wei"+UTF8.SHARP_S+"%");
1267 //        configurator.setMatchMode(MatchMode.BEGINNING);
1268 //        configurator.setDoTaxa(false);
1269 //        configurator.setDoSynonyms(false);
1270 //        configurator.setDoNamesWithoutTaxa(false);
1271 //        configurator.setDoTaxaByCommonNames(true);
1272 //
1273 //        Pager<IdentifiableEntity> pager;
1274 //
1275 //        long startMillis = System.currentTimeMillis();
1276 //        for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1277 //            pager = taxonService.findTaxaAndNames(configurator);
1278 //            if (logger.isDebugEnabled()) {
1279 //                logger.debug("[" + indx + "]" + pager.getRecords().get(0).getTitleCache());
1280 //            }
1281 //        }
1282 //        double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1283 //        logger.info("Benchmark result - [find taxon by CommonName via HQL] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1284 //    }
1285
1286     @SuppressWarnings("rawtypes")
1287     @Test
1288     @DataSet
1289     public final void benchmarkFindByCommonNameHql() throws CorruptIndexException, IOException, ParseException {
1290
1291 //        printDataSet(System.err, new String[] { "TaxonBase" });
1292
1293         createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1294
1295         IFindTaxaAndNamesConfigurator configurator = new FindTaxaAndNamesConfiguratorImpl<>();
1296         configurator.setTitleSearchString("Wei"+UTF8.SHARP_S+"%");
1297         configurator.setMatchMode(MatchMode.BEGINNING);
1298         configurator.setDoTaxa(false);
1299         configurator.setDoSynonyms(false);
1300         configurator.setDoNamesWithoutTaxa(false);
1301         configurator.setDoTaxaByCommonNames(true);
1302
1303         Pager<IdentifiableEntity> pager;
1304
1305         long startMillis = System.currentTimeMillis();
1306         for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1307             pager = taxonService.findTaxaAndNames(configurator);
1308             if (logger.isDebugEnabled()) {
1309                 logger.debug("[" + indx + "]" + pager.getRecords().get(0).getTitleCache());
1310             }
1311         }
1312         double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1313         logger.info("Benchmark result - [find taxon by CommonName via HQL] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1314     }
1315
1316     @SuppressWarnings("rawtypes")
1317     @Test
1318     @DataSet
1319     public final void benchmarkFindByCommonNameLucene() throws CorruptIndexException, IOException, ParseException {
1320
1321         createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1322
1323         refreshLuceneIndex();
1324
1325         Pager<SearchResult<TaxonBase>> pager;
1326
1327         long startMillis = System.currentTimeMillis();
1328         for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1329             pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"*", null, null, null, false, null, null, null, null);
1330             if (logger.isDebugEnabled()) {
1331                 logger.debug("[" + indx + "]" + pager.getRecords().get(0).getEntity().getTitleCache());
1332             }
1333         }
1334         double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1335         logger.info("Benchmark result - [find taxon by CommonName via lucene] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1336     }
1337
1338     /**
1339      * uncomment @Test annotation to create the dataset for this test
1340      */
1341     @Override
1342     //    @Test
1343     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class, value="BlankDataSet.xml")
1344     public final void createTestDataSet() throws FileNotFoundException {
1345
1346         Classification europeanAbiesClassification = Classification.NewInstance("European Abies");
1347         europeanAbiesClassification.setUuid(UUID.fromString(CLASSIFICATION_UUID));
1348         classificationService.save(europeanAbiesClassification);
1349
1350         Classification alternativeClassification = Classification.NewInstance("Abies alternative");
1351         alternativeClassification.setUuid(UUID.fromString(CLASSIFICATION_ALT_UUID));
1352         classificationService.save(alternativeClassification);
1353
1354         Reference sec = ReferenceFactory.newBook();
1355         sec.setTitleCache("Kohlbecker, A., Testcase standart views, 2013", true);
1356         Reference sec_sensu = ReferenceFactory.newBook();
1357         sec_sensu.setTitleCache("Komarov, V. L., Flora SSSR 29", true);
1358         referenceService.save(sec);
1359         referenceService.save(sec_sensu);
1360
1361         BotanicalName n_abies = BotanicalName.NewInstance(Rank.GENUS());
1362         n_abies.setNameCache("Abies", true);
1363         Taxon t_abies = Taxon.NewInstance(n_abies, sec);
1364         taxonService.save(t_abies);
1365
1366         BotanicalName n_abies_alba = BotanicalName.NewInstance(Rank.SPECIES());
1367         n_abies_alba.setNameCache("Abies alba", true);
1368         Taxon t_abies_alba = Taxon.NewInstance(n_abies_alba, sec);
1369         t_abies_alba.setUuid(UUID.fromString(ABIES_ALBA_UUID));
1370         taxonService.save(t_abies_alba);
1371
1372         BotanicalName n_abies_subalpina = BotanicalName.NewInstance(Rank.SPECIES());
1373         n_abies_subalpina.setNameCache("Abies subalpina", true);
1374         Synonym s_abies_subalpina = Synonym.NewInstance(n_abies_subalpina, sec);
1375         taxonService.save(s_abies_subalpina);
1376
1377         BotanicalName n_abies_balsamea = BotanicalName.NewInstance(Rank.SPECIES());
1378         n_abies_balsamea.setNameCache("Abies balsamea", true);
1379         Taxon t_abies_balsamea = Taxon.NewInstance(n_abies_balsamea, sec);
1380         t_abies_balsamea.setUuid(UUID.fromString(ABIES_BALSAMEA_UUID));
1381         t_abies_balsamea.addSynonym(s_abies_subalpina, SynonymType.SYNONYM_OF());
1382         taxonService.save(t_abies_balsamea);
1383
1384         BotanicalName n_abies_grandis = BotanicalName.NewInstance(Rank.SPECIES());
1385         n_abies_grandis.setNameCache("Abies grandis", true);
1386         Taxon t_abies_grandis = Taxon.NewInstance(n_abies_grandis, sec);
1387         taxonService.save(t_abies_grandis);
1388
1389         BotanicalName n_abies_kawakamii = BotanicalName.NewInstance(Rank.SPECIES());
1390         n_abies_kawakamii.setNameCache("Abies kawakamii", true);
1391         Taxon t_abies_kawakamii = Taxon.NewInstance(n_abies_kawakamii, sec);
1392         t_abies_kawakamii.getTitleCache();
1393         taxonService.save(t_abies_kawakamii);
1394
1395         // abies_kawakamii_sensu_komarov as missapplied name for t_abies_balsamea
1396         Taxon t_abies_kawakamii_sensu_komarov = Taxon.NewInstance(n_abies_kawakamii, sec_sensu);
1397         taxonService.save(t_abies_kawakamii_sensu_komarov);
1398         t_abies_kawakamii_sensu_komarov.addTaxonRelation(t_abies_balsamea, TaxonRelationshipType.MISAPPLIED_NAME_FOR(), null, null);
1399         taxonService.saveOrUpdate(t_abies_kawakamii_sensu_komarov);
1400
1401         BotanicalName n_abies_lasiocarpa = BotanicalName.NewInstance(Rank.SPECIES());
1402         n_abies_lasiocarpa.setNameCache("Abies lasiocarpa", true);
1403         Taxon t_abies_lasiocarpa = Taxon.NewInstance(n_abies_lasiocarpa, sec);
1404         taxonService.save(t_abies_lasiocarpa);
1405
1406         // add taxa to classifications
1407         europeanAbiesClassification.addChildTaxon(t_abies_balsamea, null, null);
1408         alternativeClassification.addChildTaxon(t_abies_lasiocarpa, null, null);
1409         classificationService.saveOrUpdate(europeanAbiesClassification);
1410         classificationService.saveOrUpdate(alternativeClassification);
1411
1412         //
1413         // Description
1414         //
1415         TaxonDescription d_abies_alba = TaxonDescription.NewInstance(t_abies_alba);
1416         TaxonDescription d_abies_balsamea = TaxonDescription.NewInstance(t_abies_balsamea);
1417
1418         d_abies_alba.setUuid(UUID.fromString(D_ABIES_ALBA_UUID));
1419         d_abies_balsamea.setUuid(UUID.fromString(D_ABIES_BALSAMEA_UUID));
1420
1421
1422         // CommonTaxonName
1423         d_abies_alba.addElement(CommonTaxonName.NewInstance("Wei"+UTF8.SHARP_S+"tanne", Language.GERMAN()));
1424         d_abies_alba.addElement(CommonTaxonName.NewInstance("silver fir", Language.ENGLISH()));
1425         d_abies_alba.addElement(Distribution
1426                 .NewInstance(
1427                         germany,
1428                         PresenceAbsenceTerm.NATIVE()));
1429         d_abies_alba.addElement(Distribution
1430                 .NewInstance(
1431                         russia,
1432                         PresenceAbsenceTerm.ABSENT()));
1433
1434         // TextData
1435         d_abies_balsamea
1436             .addElement(TextData
1437                     .NewInstance(
1438                             "Die Balsam-Tanne (Abies balsamea) ist eine Pflanzenart aus der Gattung der Tannen (Abies). Sie wÃ¤chst im nordÃ¶stlichen Nordamerika, wo sie sowohl Tief- als auch Bergland besiedelt. Sie gilt als relativ anspruchslos gegenÃ¼ber dem Standort und ist frosthart. In vielen Teilen des natÃ¼rlichen Verbreitungsgebietes stellt sie die Klimaxbaumart dar.",
1439                             Language.GERMAN(), null));
1440         d_abies_balsamea
1441         .addElement(CommonTaxonName
1442                 .NewInstance(
1443                         "Balsam-Tanne",
1444                         Language.GERMAN(), null));
1445
1446         d_abies_balsamea
1447         .addElement(Distribution
1448                 .NewInstance(
1449                         canada,
1450                         PresenceAbsenceTerm.PRESENT()));
1451
1452         d_abies_balsamea
1453         .addElement(Distribution
1454                 .NewInstance(
1455                         germany,
1456                         PresenceAbsenceTerm.NATIVE()));
1457
1458         d_abies_balsamea
1459                 .addElement(TextData
1460                         .NewInstance(
1461                                 TaxonServiceSearchTestUtf8Constants.RUSSIAN_ABIES_ALBA_LONG,
1462                                 Language.RUSSIAN(), null));
1463         d_abies_balsamea
1464         .addElement(CommonTaxonName
1465                 .NewInstance(
1466                         TaxonServiceSearchTestUtf8Constants.RUSSIAN_ABIES_ALBA_SHORT,
1467                         Language.RUSSIAN(), null));
1468         descriptionService.saveOrUpdate(d_abies_balsamea);
1469
1470         setComplete();
1471         endTransaction();
1472
1473
1474         writeDbUnitDataSetFile(new String[] {
1475             "TAXONBASE", "TAXONNAMEBASE",
1476             "TAXONRELATIONSHIP",
1477             "REFERENCE", "DESCRIPTIONELEMENTBASE", "DESCRIPTIONBASE",
1478             "AGENTBASE", "HOMOTYPICALGROUP",
1479             "CLASSIFICATION", "TAXONNODE",
1480             "LANGUAGESTRING", "DESCRIPTIONELEMENTBASE_LANGUAGESTRING",
1481             "HIBERNATE_SEQUENCES" // IMPORTANT!!!
1482             });
1483
1484     }
1485
1486     /**
1487      *
1488      */
1489     private void refreshLuceneIndex() {
1490
1491 //        commitAndStartNewTransaction(null);
1492         commit();
1493         endTransaction();
1494         indexer.purge(DefaultProgressMonitor.NewInstance());
1495         indexer.reindex(typesToIndex, DefaultProgressMonitor.NewInstance());
1496         startNewTransaction();
1497 //        commitAndStartNewTransaction(null);
1498     }
1499
1500     /**
1501      * @param numberOfNew
1502      *
1503      */
1504     private void createRandomTaxonWithCommonName(int numberOfNew) {
1505
1506         logger.debug(String.format("creating %1$s random taxan with CommonName", numberOfNew));
1507
1508         commitAndStartNewTransaction(null);
1509
1510         Reference sec = ReferenceFactory.newBook();
1511         referenceService.save(sec);
1512
1513         for (int i = numberOfNew; i < numberOfNew; i++) {
1514             RandomStringUtils.randomAlphabetic(10);
1515             String radomName = RandomStringUtils.randomAlphabetic(5) + " " + RandomStringUtils.randomAlphabetic(10);
1516             String radomCommonName = RandomStringUtils.randomAlphabetic(10);
1517
1518             BotanicalName name = BotanicalName.NewInstance(Rank.SPECIES());
1519             name.setNameCache(radomName, true);
1520             Taxon taxon = Taxon.NewInstance(name, sec);
1521             taxonService.save(taxon);
1522
1523             TaxonDescription description = TaxonDescription.NewInstance(taxon);
1524             description.addElement(CommonTaxonName.NewInstance(radomCommonName, Language.GERMAN()));
1525             descriptionService.save(description);
1526         }
1527
1528         commitAndStartNewTransaction(null);
1529     }
1530
1531     private <T extends CdmBase> void logSearchResults(Pager<SearchResult<T>> pager, Level level, String[] docFields){
1532         if(level == null){
1533             level = Level.DEBUG;
1534         }
1535         if(logger.isEnabledFor(level)){
1536             StringBuilder b = new StringBuilder();
1537             b.append("\n");
1538             int i = 0;
1539             for(SearchResult sr : pager.getRecords()){
1540
1541                 b.append(" ").append(i++).append(" - ");
1542                 b.append("score:").append(sr.getScore()).append(", ");
1543
1544                 if(docFields != null){
1545                     b.append("docs : ");
1546                     for(Document doc : sr.getDocs()) {
1547                         b.append("<");
1548                         for(String f : docFields){
1549                             b.append(f).append(":").append(Arrays.toString(doc.getValues(f)));
1550                         }
1551                         b.append(">");
1552                     }
1553                 }
1554
1555                 CdmBase entity = sr.getEntity();
1556                 if(entity == null){
1557                     b.append("NULL");
1558                 } else {
1559                     b.append(entity.getClass().getSimpleName()).
1560                         append(" [").append(entity.getId()).
1561                         append(" | ").append(entity.getUuid()).append("] : ").
1562                         append(entity.toString());
1563
1564                 }
1565                 b.append("\n");
1566             }
1567             logger.log(level, b);
1568         }
1569     }
1570
1571 }