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.assertNotNull;
  13
  14 import java.io.FileNotFoundException;
  15 import java.io.IOException;
  16 import java.util.ArrayList;
  17 import java.util.Arrays;
  18 import java.util.EnumSet;
  19 import java.util.HashSet;
  20 import java.util.List;
  21 import java.util.Map;
  22 import java.util.Set;
  23 import java.util.UUID;
  24
  25 import org.apache.commons.lang.RandomStringUtils;
  26 import org.apache.log4j.Level;
  27 import org.apache.log4j.Logger;
  28 import org.apache.lucene.document.Document;
  29 import org.junit.Assert;
  30 import org.junit.Before;
  31 import org.junit.Ignore;
  32 import org.junit.Test;
  33 import org.unitils.dbunit.annotation.DataSet;
  34 import org.unitils.spring.annotation.SpringBeanByType;
  35
  36 import eu.etaxonomy.cdm.api.service.config.FindTaxaAndNamesConfiguratorImpl;
  37 import eu.etaxonomy.cdm.api.service.config.IFindTaxaAndNamesConfigurator;
  38 import eu.etaxonomy.cdm.api.service.pager.Pager;
  39 import eu.etaxonomy.cdm.api.service.search.ICdmMassIndexer;
  40 import eu.etaxonomy.cdm.api.service.search.LuceneMultiSearchException;
  41 import eu.etaxonomy.cdm.api.service.search.LuceneParseException;
  42 import eu.etaxonomy.cdm.api.service.search.SearchResult;
  43 import eu.etaxonomy.cdm.common.UTF8;
  44 import eu.etaxonomy.cdm.common.monitor.DefaultProgressMonitor;
  45 import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper;
  46 import eu.etaxonomy.cdm.model.common.CdmBase;
  47 import eu.etaxonomy.cdm.model.common.DefinedTermBase;
  48 import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
  49 import eu.etaxonomy.cdm.model.common.Language;
  50 import eu.etaxonomy.cdm.model.description.CategoricalData;
  51 import eu.etaxonomy.cdm.model.description.CommonTaxonName;
  52 import eu.etaxonomy.cdm.model.description.DescriptionBase;
  53 import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
  54 import eu.etaxonomy.cdm.model.description.Distribution;
  55 import eu.etaxonomy.cdm.model.description.Feature;
  56 import eu.etaxonomy.cdm.model.description.PresenceAbsenceTerm;
  57 import eu.etaxonomy.cdm.model.description.State;
  58 import eu.etaxonomy.cdm.model.description.StateData;
  59 import eu.etaxonomy.cdm.model.description.TaxonDescription;
  60 import eu.etaxonomy.cdm.model.description.TextData;
  61 import eu.etaxonomy.cdm.model.location.Country;
  62 import eu.etaxonomy.cdm.model.location.NamedArea;
  63 import eu.etaxonomy.cdm.model.name.IBotanicalName;
  64 import eu.etaxonomy.cdm.model.name.Rank;
  65 import eu.etaxonomy.cdm.model.name.TaxonName;
  66 import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
  67 import eu.etaxonomy.cdm.model.reference.Reference;
  68 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  69 import eu.etaxonomy.cdm.model.taxon.Classification;
  70 import eu.etaxonomy.cdm.model.taxon.Synonym;
  71 import eu.etaxonomy.cdm.model.taxon.SynonymType;
  72 import eu.etaxonomy.cdm.model.taxon.Taxon;
  73 import eu.etaxonomy.cdm.model.taxon.TaxonBase;
  74 import eu.etaxonomy.cdm.model.taxon.TaxonNode;
  75 import eu.etaxonomy.cdm.model.taxon.TaxonRelationship;
  76 import eu.etaxonomy.cdm.model.taxon.TaxonRelationshipType;
  77 import eu.etaxonomy.cdm.persistence.query.MatchMode;
  78 import eu.etaxonomy.cdm.persistence.query.OrderHint;
  79 import eu.etaxonomy.cdm.test.integration.CdmTransactionalIntegrationTest;
  80 import eu.etaxonomy.cdm.test.unitils.CleanSweepInsertLoadStrategy;
  81
  82 /**
  83  * @author a.babadshanjan, a.kohlbecker
  84  * @since 04.02.2009
  85  */
  86 public class TaxonServiceSearchTest extends CdmTransactionalIntegrationTest {
  87
  88     private static Logger logger = Logger.getLogger(TaxonServiceSearchTest.class);
  89
  90     private static final UUID ABIES_BALSAMEA_UUID = UUID.fromString("f65d47bd-4f49-4ab1-bc4a-bc4551eaa1a8");
  91     private static final UUID ABIES_ALBA_UUID = UUID.fromString("7dbd5810-a3e5-44b6-b563-25152b8867f4");
  92     private static final UUID CLASSIFICATION_UUID = UUID.fromString("2a5ceebb-4830-4524-b330-78461bf8cb6b");
  93     private static final UUID CLASSIFICATION_ALT_UUID = UUID.fromString("d7c741e3-ae9e-4a7d-a566-9e3a7a0b51ce");
  94     private static final UUID D_ABIES_BALSAMEA_UUID = UUID.fromString("900108d8-e6ce-495e-b32e-7aad3099135e");
  95     private static final UUID D_ABIES_ALBA_UUID = UUID.fromString("ec8bba03-d993-4c85-8472-18b14942464b");
  96     private static final UUID D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID = UUID.fromString("e9d8c2fd-6409-46d5-9c2e-14a2bbb1b2b1");
  97     private static final UUID ABIES_SUBALPINA_UUID = UUID.fromString("9fee273c-c819-4f1f-913a-cd910465df51");
  98     private static final UUID ABIES_LASIOCARPA_UUID = UUID.fromString("9ce1fecf-c1ad-4127-be01-85d5d9f847ce");
  99
 100     private static final int NUM_OF_NEW_RADOM_ENTITIES = 1000;
 101
 102     private boolean includeUnpublished = true;
 103
 104
 105
 106     @SpringBeanByType
 107     private ITaxonService taxonService;
 108     @SpringBeanByType
 109     private ITermService termService;
 110     @SpringBeanByType
 111     private IClassificationService classificationService;
 112     @SpringBeanByType
 113     private IReferenceService referenceService;
 114     @SpringBeanByType
 115     private IDescriptionService descriptionService;
 116     @SpringBeanByType
 117     private INameService nameService;
 118     @SpringBeanByType
 119     private ICdmMassIndexer indexer;
 120
 121     @SpringBeanByType
 122     private ITaxonNodeService nodeService;
 123
 124     private static final int BENCHMARK_ROUNDS = 300;
 125
 126     private Set<Class<? extends CdmBase>> typesToIndex = null;
 127
 128     private NamedArea germany;
 129     private NamedArea france ;
 130     private NamedArea russia ;
 131     private NamedArea canada ;
 132
 133     @Before
 134     public void setUp() throws Exception {
 135         typesToIndex = new HashSet<>();
 136         typesToIndex.add(DescriptionElementBase.class);
 137         typesToIndex.add(TaxonBase.class);
 138         typesToIndex.add(TaxonRelationship.class);
 139
 140         germany =  Country.GERMANY();
 141         france = Country.FRANCEFRENCHREPUBLIC();
 142         russia = Country.RUSSIANFEDERATION();
 143         canada = Country.CANADA();
 144
 145         includeUnpublished = true;
 146     }
 147
 148     @Test
 149     public void testDbUnitUsageTest() throws Exception {
 150         assertNotNull("taxonService should exist", taxonService);
 151         assertNotNull("nameService should exist", nameService);
 152     }
 153
 154
 155     @SuppressWarnings("rawtypes")
 156     @Test
 157     @DataSet
 158     public final void testPurgeAndReindex() throws IOException, LuceneParseException {
 159
 160         refreshLuceneIndex();
 161         TaxonNode subtree = null;
 162
 163         Pager<SearchResult<TaxonBase>> pager = taxonService.findByFullText(null, "Abies", null, subtree,
 164                 includeUnpublished, null, true, null, null, null, null); // --> 8
 165         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 166
 167         indexer.purge(null);
 168         commitAndStartNewTransaction(null);
 169
 170         pager = taxonService.findByFullText(null, "Abies", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 0
 171         Assert.assertEquals("Expecting no entities since the index has been purged", 0, pager.getCount().intValue());
 172
 173         indexer.reindex(indexer.indexedClasses(), null);
 174         commitAndStartNewTransaction(null);
 175
 176         pager = taxonService.findByFullText(null, "Abies", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 8
 177         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 178     }
 179
 180
 181     @SuppressWarnings("rawtypes")
 182     @Test
 183     @DataSet
 184     public final void testFindByDescriptionElementFullText_CommonName() throws IOException,
 185             LuceneParseException {
 186
 187         TaxonNode subtree = null;
 188         refreshLuceneIndex();
 189
 190         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne",
 191                 null, subtree, null, null,
 192                 false, null, null, null, null);
 193         Assert.assertEquals("Expecting one entity when searching for CommonTaxonName", 1,
 194                 pager.getCount().intValue());
 195
 196         // the description containing the Nulltanne has no taxon attached,
 197         // taxon.id = null
 198         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Nulltanne", null, subtree, null, null,
 199                 false, null, null, null, null);
 200         Assert.assertEquals("Expecting no entity when searching for 'Nulltanne' ", 0, pager.getCount().intValue());
 201
 202         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne", null, subtree, null,
 203                 Arrays.asList(new Language[] { Language.GERMAN() }), false, null, null, null, null);
 204         Assert.assertEquals("Expecting one entity when searching in German", 1, pager.getCount().intValue());
 205
 206         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"tanne", null, subtree, null,
 207                 Arrays.asList(new Language[] { Language.RUSSIAN() }), false, null, null, null, null);
 208         Assert.assertEquals("Expecting no entity when searching in Russian", 0, pager.getCount().intValue());
 209
 210     }
 211
 212     @SuppressWarnings("rawtypes")
 213     @Test
 214     @DataSet
 215     public final void testFindByDescriptionElementFullText_Distribution() throws IOException, LuceneParseException {
 216
 217         refreshLuceneIndex();
 218         TaxonNode subtree = null;
 219
 220         // by Area
 221         Pager<SearchResult<TaxonBase>>  pager = taxonService.findByDescriptionElementFullText(null, "Canada", null, subtree, null, null, false, null, null, null, null);
 222         Assert.assertEquals("Expecting one entity when searching for arae 'Canada'", 1, pager.getCount().intValue());
 223         // by Status
 224         pager = taxonService.findByDescriptionElementFullText(null, "present", null, subtree, null, null, false, null, null, null, null);
 225         Assert.assertEquals("Expecting one entity when searching for status 'present'", 1, pager.getCount().intValue());
 226     }
 227
 228     @SuppressWarnings("rawtypes")
 229     @Test
 230     @DataSet
 231     public final void testFindByDescriptionElementFullText_wildcard() throws IOException, LuceneParseException {
 232
 233         refreshLuceneIndex();
 234         TaxonNode subtree = null;
 235
 236         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"*", null, subtree, null, null, false, null, null, null, null);
 237         Assert.assertEquals("Expecting one entity when searching for CommonTaxonName", 1, pager.getCount().intValue());
 238     }
 239
 240     /**
 241      * Regression test for #3113 (hibernate search: wildcard query can cause BooleanQuery$TooManyClauses: maxClauseCount is set to 1024)
 242      *
 243      * @throws IOException
 244      * @throws LuceneParseException
 245      */
 246     @SuppressWarnings("rawtypes")
 247     @Test
 248     @DataSet
 249     public final void testFindByDescriptionElementFullText_TooManyClauses() throws IOException, LuceneParseException {
 250
 251         TaxonNode subtree = null;
 252
 253         // generate 1024 terms to reproduce the bug
 254         TaxonDescription description = (TaxonDescription) descriptionService.find(D_ABIES_ALBA_UUID);
 255         Set<String> uniqueRandomStrs = new HashSet<>(1024);
 256         while(uniqueRandomStrs.size() < 1024){
 257             uniqueRandomStrs.add(RandomStringUtils.random(10, true, false));
 258         }
 259         for(String rndStr: uniqueRandomStrs){
 260             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 261         }
 262         descriptionService.saveOrUpdate(description);
 263         commitAndStartNewTransaction(null);
 264
 265         refreshLuceneIndex();
 266
 267         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, subtree, null, null, false, null, null, null, null);
 268         Assert.assertEquals("Expecting all 1024 entities grouped into one SearchResult item when searching for Rot*", 1, pager.getCount().intValue());
 269     }
 270
 271     /**
 272      * Regression test for #3116 (fulltext search: always only one page of results)
 273      *
 274      * @throws IOException
 275      * @throws LuceneParseException
 276      */
 277     @SuppressWarnings("rawtypes")
 278     @Test
 279     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 280     public final void testFullText_Paging() throws IOException, LuceneParseException {
 281
 282         TaxonNode subtree = null;
 283         Reference sec = ReferenceFactory.newDatabase();
 284         referenceService.save(sec);
 285
 286         Set<String> uniqueRandomStrs = new HashSet<>(1024);
 287         int numOfItems = 100;
 288         while(uniqueRandomStrs.size() < numOfItems){
 289             uniqueRandomStrs.add(RandomStringUtils.random(5, true, false));
 290         }
 291
 292         for(String rndStr: uniqueRandomStrs){
 293
 294             Taxon taxon = Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(Rank.SERIES()), sec);
 295             taxon.setTitleCache("Tax" + rndStr, true);
 296             taxonService.save(taxon);
 297
 298             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 299             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 300             descriptionService.saveOrUpdate(description);
 301         }
 302
 303         commitAndStartNewTransaction(new String[]{"TAXONBASE"});
 304         refreshLuceneIndex();
 305
 306         int pageSize = 10;
 307
 308         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, subtree, null, null, false, pageSize, null, null, null);
 309         Assert.assertEquals("unexpeted number of pages", Integer.valueOf(numOfItems / pageSize), pager.getPagesAvailable());
 310         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, subtree, null, null, false, pageSize, 9, null, null);
 311         Assert.assertNotNull("last page must have records", pager.getRecords());
 312         Assert.assertNotNull("last item on last page must exist", pager.getRecords().get(0));
 313         pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, subtree, null, null, false, pageSize, 10, null, null);
 314         Assert.assertNotNull("last page + 1 must not have any records", pager.getRecords());
 315     }
 316
 317     /**
 318      * test for max score and sort by score of hit groups
 319      * with all matches per taxon in a single TextData  element
 320      * see {@link #testFullText_ScoreAndOrder_2()} for the complement
 321      * test with matches in multiple TextData per taxon
 322      *
 323      * @throws IOException
 324      * @throws LuceneParseException
 325      */
 326     @SuppressWarnings("rawtypes")
 327     @Test
 328     @DataSet
 329     @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
 330     public final void testFullText_ScoreAndOrder_1() throws IOException, LuceneParseException {
 331
 332         TaxonNode subtree = null;
 333         int numOfTaxa = 3;
 334
 335         UUID[] taxonUuids = new UUID[numOfTaxa];
 336         StringBuilder text = new StringBuilder();
 337
 338         for(int i = 0; i < numOfTaxa; i++){
 339
 340             Taxon taxon = Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(null), null);
 341             taxon.setTitleCache("Taxon_" + i, true);
 342             taxonUuids[i] = taxon.getUuid();
 343             taxonService.save(taxon);
 344
 345             text.append(" ").append("Rot");
 346             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 347             description.addElement(TextData.NewInstance(text.toString(), Language.DEFAULT(), null));
 348             descriptionService.saveOrUpdate(description);
 349         }
 350
 351         commitAndStartNewTransaction(null);
 352         refreshLuceneIndex();
 353
 354         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Rot", null, subtree, null, null, false, null, null, null, null);
 355         for(int i = 0; i < numOfTaxa; i++){
 356             Assert.assertEquals("taxa should be orderd by relevance (= score)", taxonUuids[numOfTaxa - i - 1], pager.getRecords().get(i).getEntity().getUuid());
 357         }
 358         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);
 359     }
 360
 361     /**
 362      * test for max score and sort by score of hit groups
 363      * with all matches per taxon in a multiple TextData elements
 364      * see {@link #testFullText_ScoreAndOrder_1()} for the complement
 365      * test with matches in a single TextData per taxon
 366      *
 367      * @throws IOException
 368      * @throws LuceneParseException
 369      */
 370     @SuppressWarnings("rawtypes")
 371     @Test
 372     @DataSet
 373     @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
 374     public final void testFullText_ScoreAndOrder_2() throws IOException, LuceneParseException {
 375
 376         TaxonNode subtree = null;
 377         int numOfTaxa = 3;
 378
 379         UUID[] taxonUuids = new UUID[numOfTaxa];
 380
 381         for(int i = 0; i < numOfTaxa; i++){
 382
 383             Taxon taxon = Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(null), null);
 384             taxon.setTitleCache("Taxon_" + i, true);
 385             taxonUuids[i] = taxon.getUuid();
 386             taxonService.save(taxon);
 387
 388             TaxonDescription description = TaxonDescription.NewInstance(taxon);
 389             for(int k = 0; k < i; k++){
 390                 description.addElement(TextData.NewInstance("Rot", Language.DEFAULT(), null));
 391             }
 392             descriptionService.saveOrUpdate(description);
 393         }
 394
 395         commitAndStartNewTransaction(null);
 396         refreshLuceneIndex();
 397
 398         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Rot", null, subtree, null, null, false, null, null, null, null);
 399         for(int i = 0; i < numOfTaxa; i++){
 400             Assert.assertEquals("taxa should be orderd by relevance (= score)", taxonUuids[numOfTaxa - i - 1], pager.getRecords().get(i).getEntity().getUuid());
 401         }
 402         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);
 403     }
 404
 405
 406     /**
 407      * @throws IOException
 408      * @throws LuceneParseException
 409      * @throws LuceneMultiSearchException
 410      */
 411     @Test
 412     @DataSet
 413     public final void testFullText_Grouping() throws IOException, LuceneParseException, LuceneMultiSearchException {
 414
 415         TaxonNode subtree = null;
 416         TaxonDescription description = (TaxonDescription) descriptionService.find(D_ABIES_ALBA_UUID);
 417         Set<String> uniqueRandomStrs = new HashSet<>(1024);
 418         int numOfItems = 100;
 419         while(uniqueRandomStrs.size() < numOfItems){
 420             uniqueRandomStrs.add(RandomStringUtils.random(5, true, false));
 421         }
 422         for(String rndStr: uniqueRandomStrs){
 423             description.addElement(CommonTaxonName.NewInstance("Rot" + rndStr, Language.DEFAULT()));
 424         }
 425         descriptionService.saveOrUpdate(description);
 426
 427         commitAndStartNewTransaction(new String[]{"DESCRIPTIONELEMENTBASE"});
 428
 429         refreshLuceneIndex();
 430
 431         int pageSize = 10;
 432
 433          boolean highlightFragments = true;
 434
 435         // test with findByDescriptionElementFullText
 436         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Rot*", null, subtree, null, null, highlightFragments, pageSize, null, null, null);
 437         logFreeTextSearchResults(pager, Level.DEBUG, null);
 438         Assert.assertEquals("All matches should be grouped into a single SearchResult element", 1, pager.getRecords().size());
 439         Assert.assertEquals("The count property of the pager must be set correctly", 1, pager.getCount().intValue());
 440         Map<String, String[]> highlightMap = pager.getRecords().get(0).getFieldHighlightMap();
 441         // maxDocsPerGroup is defined in LuceneSearch and defaults to 10
 442         int maxDocsPerGroup = 10;
 443         Assert.assertEquals("expecting 10 highlighted fragments of field 'name'", maxDocsPerGroup, highlightMap.get("name").length);
 444
 445         // test with findByEverythingFullText
 446         pager = taxonService.findByEverythingFullText( "Rot*", null, subtree, includeUnpublished, null, highlightFragments, pageSize, null, null, null);
 447         logFreeTextSearchResults(pager, Level.DEBUG, null);
 448         Assert.assertEquals("All matches should be grouped into a single SearchResult element", 1, pager.getRecords().size());
 449         Assert.assertEquals("The count property of the pager must be set correctly", 1, pager.getCount().intValue());
 450         highlightMap = pager.getRecords().get(0).getFieldHighlightMap();
 451         // maxDocsPerGroup is defined in LuceneSearch and defaults to 10
 452         maxDocsPerGroup = 10;
 453         Assert.assertEquals("expecting 10 highlighted fragments of field 'name'", maxDocsPerGroup, highlightMap.get("name").length);
 454
 455     }
 456
 457     @SuppressWarnings("rawtypes")
 458     @Test
 459     @DataSet
 460     public final void testFindByDescriptionElementFullText_TextData() throws IOException, LuceneParseException {
 461
 462         refreshLuceneIndex();
 463         TaxonNode subtree = null;
 464
 465         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Abies", null, subtree, null, null, false, null, null, null, null);
 466         logFreeTextSearchResults(pager, Level.DEBUG, null);
 467         Assert.assertEquals("Expecting one entity when searching for any TextData", 1, pager.getCount().intValue());
 468         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 469         Assert.assertTrue("Expecting two docs, one for RUSSIAN and one for GERMAN", pager.getRecords().get(0).getDocs().size() == 2);
 470         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getDocs().iterator().next().get("inDescription.taxon.titleCache"));
 471
 472
 473         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, subtree, null, null, false, null, null, null, null);
 474         Assert.assertEquals("Expecting one entity when searching for any type", 1, pager.getCount().intValue());
 475
 476         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, subtree, Arrays.asList(new Feature[]{Feature.UNKNOWN()}), null, false, null, null, null, null);
 477         Assert.assertEquals("Expecting one entity when searching for any type and for Feature DESCRIPTION", 1, pager.getCount().intValue());
 478
 479         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, subtree, Arrays.asList(new Feature[]{Feature.CHROMOSOME_NUMBER()}), null, false, null, null, null, null);
 480         Assert.assertEquals("Expecting no entity when searching for any type and for Feature CHROMOSOME_NUMBER", 0, pager.getCount().intValue());
 481
 482         pager = taxonService.findByDescriptionElementFullText(null, "Abies", null, subtree, Arrays.asList(new Feature[]{Feature.CHROMOSOME_NUMBER(), Feature.UNKNOWN()}), null, false, null, null, null, null);
 483         Assert.assertEquals("Expecting no entity when searching for any type and for Feature DESCRIPTION or CHROMOSOME_NUMBER", 1, pager.getCount().intValue());
 484
 485         pager = taxonService.findByDescriptionElementFullText(Distribution.class, "Abies", null, subtree, null, null, false, null, null, null, null);
 486         Assert.assertEquals("Expecting no entity when searching for Distribution", 0, pager.getCount().intValue());
 487
 488         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, subtree, null, Arrays.asList(new Language[]{}), false, null, null, null, null);
 489         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 490         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 491
 492         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, subtree, null, Arrays.asList(new Language[]{Language.RUSSIAN()}), false, null, null, null, null);
 493         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 494         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 495
 496         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Бальзам", null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 497         Assert.assertEquals("Expecting no entity", 0, pager.getCount().intValue());
 498
 499         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 500         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 501         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 502     }
 503
 504     @SuppressWarnings("rawtypes")
 505     @Test
 506     @DataSet
 507     public final void testFindByDescriptionElementFullText_MultipleWords() throws IOException, LuceneParseException {
 508
 509         refreshLuceneIndex();
 510         TaxonNode subtree = null;
 511
 512         // Pflanzenart aus der Gattung der Tannen
 513         long start = System.currentTimeMillis();
 514
 515         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Tannen", null, subtree, null, null, false, null, null, null, null);
 516         Assert.assertEquals("OR search : Expecting one entity", 1, pager.getCount().intValue());
 517
 518         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Wespen", null, subtree, null, null, false, null, null, null, null);
 519         Assert.assertEquals("OR search : Expecting one entity", 1, pager.getCount().intValue());
 520
 521         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Tannen", null, subtree, null, null, false, null, null, null, null);
 522         Assert.assertEquals("AND search : Expecting one entity", 1, pager.getCount().intValue());
 523
 524         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Wespen", null, subtree, null, null, false, null, null, null, null);
 525         Assert.assertEquals("AND search : Expecting no entity", 0, pager.getCount().intValue());
 526
 527         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Tannen\"", null, subtree, null, null, false, null, null, null, null);
 528         Assert.assertEquals("Phrase search : Expecting one entity", 1, pager.getCount().intValue());
 529
 530         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Wespen\"", null, subtree, null, null, false, null, null, null, null);
 531         Assert.assertEquals("Phrase search : Expecting one entity", 0, pager.getCount().intValue());
 532
 533         logger.info("testFindByDescriptionElementFullText_MultipleWords() duration: " + (System.currentTimeMillis() - start) + "ms");
 534
 535     }
 536
 537
 538     @SuppressWarnings("rawtypes")
 539     @Test
 540     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 541     public final void testFindByDescriptionElementFullText_modify_DescriptionElement() throws IOException, LuceneParseException {
 542
 543         refreshLuceneIndex();
 544         TaxonNode subtree = null;
 545
 546         //
 547         // modify the DescriptionElement
 548         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 549         Assert.assertTrue("Search did not return any results", pager.getRecords().size() > 0);
 550         Assert.assertTrue("Expecting only one doc", pager.getRecords().get(0).getDocs().size() == 1);
 551         Document indexDocument = pager.getRecords().get(0).getDocs().iterator().next();
 552         String[] descriptionElementUuidStr = indexDocument.getValues("uuid");
 553         String[] inDescriptionUuidStr = indexDocument.getValues("inDescription.uuid");
 554         // is only one uuid!
 555         DescriptionElementBase textData = descriptionService.getDescriptionElementByUuid(UUID.fromString(descriptionElementUuidStr[0]));
 556
 557         ((TextData)textData).removeText(Language.GERMAN());
 558         ((TextData)textData).putText(Language.SPANISH_CASTILIAN(), "abeto bals"+UTF8.SMALL_A_ACUTE+"mico");
 559
 560         descriptionService.saveDescriptionElement(textData);
 561         commitAndStartNewTransaction(null);
 562 //        printDataSet(System.out, new String[] {
 563 //                "DESCRIPTIONELEMENTBASE", "LANGUAGESTRING", "DESCRIPTIONELEMENTBASE_LANGUAGESTRING" }
 564 //        );
 565
 566         //
 567         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN(), Language.RUSSIAN()}), false, null, null, null, null);
 568         Assert.assertEquals("The german 'Balsam-Tanne' TextData should no longer be indexed", 0, pager.getCount().intValue());
 569         pager = taxonService.findByDescriptionElementFullText(TextData.class, "abeto", null, subtree, null, Arrays.asList(new Language[]{Language.SPANISH_CASTILIAN()}), false, null, null, null, null);
 570         Assert.assertEquals("expecting to find the SPANISH_CASTILIAN 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico'", 1, pager.getCount().intValue());
 571         pager = taxonService.findByDescriptionElementFullText(TextData.class, "bals"+UTF8.SMALL_A_ACUTE+"mico", null, subtree, null, null, false, null, null, null, null);
 572         Assert.assertEquals("expecting to find the SPANISH_CASTILIAN 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico'", 1, pager.getCount().intValue());
 573
 574         //
 575         // modify the DescriptionElement via the Description object
 576         DescriptionBase<?> description = descriptionService.find(UUID.fromString(inDescriptionUuidStr[0]));
 577         Set<DescriptionElementBase> elements = description.getElements();
 578         for( DescriptionElementBase elm : elements){
 579             if(elm.getUuid().toString().equals(descriptionElementUuidStr[0])){
 580                 ((TextData)elm).removeText(Language.SPANISH_CASTILIAN());
 581                 ((TextData)elm).putText(Language.POLISH(), "Jod"+UTF8.POLISH_L+"a balsamiczna");
 582             }
 583         }
 584         descriptionService.saveOrUpdate(description);
 585         commitAndStartNewTransaction(null);
 586         pager = taxonService.findByDescriptionElementFullText(TextData.class, "abeto", null, subtree, null, Arrays.asList(new Language[]{Language.SPANISH_CASTILIAN()}), false, null, null, null, null);
 587         Assert.assertEquals("The spanish 'abeto bals"+UTF8.SMALL_A_ACUTE+"mico' TextData should no longer be indexed", 0, pager.getCount().intValue());
 588         pager = taxonService.findByDescriptionElementFullText(TextData.class, "balsamiczna", null, subtree, null, Arrays.asList(new Language[]{Language.POLISH()}), false, null, null, null, null);
 589         Assert.assertEquals("expecting to find the POLISH 'Jod"+UTF8.POLISH_L+"a balsamiczna'", 1, pager.getCount().intValue());
 590     }
 591
 592     @SuppressWarnings("rawtypes")
 593     @Test
 594     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 595     public final void testFindByDescriptionElementFullText_modify_Taxon() throws IOException, LuceneParseException {
 596
 597         refreshLuceneIndex();
 598         TaxonNode subtree = null;
 599
 600         Taxon t_abies_balsamea = (Taxon)taxonService.find(ABIES_BALSAMEA_UUID);
 601         TaxonDescription d_abies_balsamea = (TaxonDescription)descriptionService.find(D_ABIES_BALSAMEA_UUID);
 602
 603         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne",
 604                 null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 605         Assert.assertEquals("expecting to find the GERMAN 'Balsam-Tanne'", 1, pager.getCount().intValue());
 606
 607         // exchange the Taxon with another one via the Taxon object
 608         // 1.) remove existing description:
 609         t_abies_balsamea.removeDescription(d_abies_balsamea);
 610
 611         taxonService.saveOrUpdate(t_abies_balsamea);
 612         commitAndStartNewTransaction(null);
 613
 614         t_abies_balsamea = (Taxon)taxonService.find(t_abies_balsamea.getUuid());
 615
 616         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne",
 617                 null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 618         Assert.assertEquals("'Balsam-Tanne' should no longer be found", 0, pager.getCount().intValue());
 619
 620         // 2.) create new description and add to taxon:
 621         TaxonDescription d_abies_balsamea_new = TaxonDescription.NewInstance();
 622         d_abies_balsamea_new
 623                 .addElement(TextData
 624                         .NewInstance(
 625                                 "Die Balsamtanne ist mit bis zu 30 m Höhe ein mittelgro"+UTF8.SHARP_S+"er Baum und kann bis zu 200 Jahre alt werden",
 626                                 Language.GERMAN(), null));
 627         t_abies_balsamea.addDescription(d_abies_balsamea_new);
 628         // set authorshipCache to null to avoid validation exception,
 629         // this is maybe not needed in future,  see ticket #3344
 630         IBotanicalName abies_balsamea = CdmBase.deproxy(t_abies_balsamea.getName(), TaxonName.class);
 631         abies_balsamea.setAuthorshipCache(null);
 632         printDataSet(System.err, new String[] {"LANGUAGESTRING_AUD"});
 633         taxonService.saveOrUpdate(t_abies_balsamea);
 634         commitAndStartNewTransaction(null);
 635
 636 //        printDataSet(System.out, new String[] {
 637 //                "DESCRIPTIONBASE"
 638 //        });
 639
 640         pager = taxonService.findByDescriptionElementFullText(TextData.class, "mittelgro"+UTF8.SHARP_S+"er Baum",
 641                 null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 642         Assert.assertEquals("the taxon should be found via the new Description", 1, pager.getCount().intValue());
 643     }
 644
 645     @SuppressWarnings("rawtypes")
 646     @Test
 647     @DataSet
 648     public final void testFindByDescriptionElementFullText_modify_Classification() throws IOException, LuceneParseException {
 649
 650         refreshLuceneIndex();
 651         TaxonNode subtree = null;
 652
 653         // put taxon into other classification, new taxon node
 654         Classification classification = classificationService.find(CLASSIFICATION_UUID);
 655         Classification alternateClassification = classificationService.find(CLASSIFICATION_ALT_UUID);
 656
 657         // TODO: why is the test failing when the childNode is already retrieved here, and not after the following four lines?
 658         //TaxonNode childNode = classification.getChildNodes().iterator().next();
 659
 660         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", null, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 661         Assert.assertEquals("expecting to find the GERMAN 'Balsam-Tanne' even if filtering by classification", 1, pager.getCount().intValue());
 662         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne", alternateClassification, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 663         Assert.assertEquals("GERMAN 'Balsam-Tanne' should NOT be found in other classification", 0, pager.getCount().intValue());
 664
 665         // check for the right taxon node
 666         TaxonNode childNode = classification.getChildNodes().iterator().next();
 667         Assert.assertEquals("expecting Abies balsamea sec.", childNode.getTaxon().getUuid(), ABIES_BALSAMEA_UUID);
 668         Assert.assertEquals("expecting default classification", childNode.getClassification().getUuid(), CLASSIFICATION_UUID);
 669
 670         // moving the taxon around, the rootnode is only a proxy
 671         alternateClassification.setRootNode(HibernateProxyHelper.deproxy(alternateClassification.getRootNode(), TaxonNode.class));
 672         alternateClassification.addChildNode(childNode, null, null);
 673
 674         classificationService.saveOrUpdate(alternateClassification);
 675         commitAndStartNewTransaction(null);
 676
 677 //        printDataSet(System.out, new String[] {
 678 //            "TAXONBASE", "TAXONNODE", "CLASSIFICATION"
 679 //        });
 680
 681         // reload classification
 682         classification = classificationService.find(CLASSIFICATION_UUID);
 683         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne",
 684                 alternateClassification, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 685         Assert.assertEquals("GERMAN 'Balsam-Tanne' should now be found in other classification", 1, pager.getCount().intValue());
 686
 687         classification.getChildNodes().clear();
 688         classificationService.saveOrUpdate(classification);
 689         commitAndStartNewTransaction(null);
 690
 691         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Balsam-Tanne",
 692                 classification, subtree, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
 693         Assert.assertEquals("Now the GERMAN 'Balsam-Tanne' should NOT be found in original classification", 0, pager.getCount().intValue());
 694
 695     }
 696
 697     @SuppressWarnings("rawtypes")
 698     @Test
 699     @DataSet
 700     public final void testFindByDescriptionElementFullText_CategoricalData() throws IOException, LuceneParseException {
 701
 702         TaxonNode subtree = null;
 703         // add CategoricalData
 704         DescriptionBase d_abies_balsamea = descriptionService.find(D_ABIES_BALSAMEA_UUID);
 705         // Categorical data
 706         CategoricalData cdata = CategoricalData.NewInstance();
 707         cdata.setFeature(Feature.DESCRIPTION());
 708         State state = State.NewInstance("green", "green", "gn");
 709
 710         StateData statedata = StateData.NewInstance(state);
 711         statedata.putModifyingText(Language.ENGLISH(), "always, even during winter");
 712         cdata.addStateData(statedata);
 713         d_abies_balsamea.addElement(cdata);
 714
 715         UUID termUUID = termService.save(state).getUuid();
 716         descriptionService.save(d_abies_balsamea);
 717
 718         commitAndStartNewTransaction(null);
 719
 720 //        printDataSet(System.out, new String[] {
 721 //                 "STATEDATA", "STATEDATA_DEFINEDTERMBASE", "STATEDATA_LANGUAGESTRING", "LANGUAGESTRING"});
 722
 723         refreshLuceneIndex();
 724
 725         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(CategoricalData.class, "green", null, subtree, null, null, false, null, null, null, null);
 726         Assert.assertEquals("Expecting one entity", 1, pager.getCount().intValue());
 727         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getEntity().getTitleCache());
 728         Assert.assertTrue("Expecting only one doc", pager.getRecords().get(0).getDocs().size() == 1);
 729         Assert.assertEquals("Abies balsamea sec. Kohlbecker, A., Testcase standart views, 2013", pager.getRecords().get(0).getDocs().iterator().next().get("inDescription.taxon.titleCache"));
 730
 731
 732         //TODO modify the StateData
 733         TaxonBase taxon = pager.getRecords().get(0).getEntity();
 734
 735         String newName = "Quercus robur";
 736         taxon.setTitleCache(newName + " sec. ", true);
 737
 738         taxonService.saveOrUpdate(taxon);
 739         commitAndStartNewTransaction(null);
 740
 741         taxon = taxonService.find(taxon.getUuid());
 742         Assert.assertEquals(newName + " sec. ", taxon.getTitleCache());
 743         DefinedTermBase term = termService.find(termUUID);
 744
 745         termService.delete(term);
 746
 747     }
 748
 749     @SuppressWarnings("rawtypes")
 750     @Test
 751     @DataSet
 752     public final void testFindByDescriptionElementFullText_Highlighting() throws IOException, LuceneParseException {
 753
 754         refreshLuceneIndex();
 755         TaxonNode subtree = null;
 756
 757         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDescriptionElementFullText(TextData.class, "Abies", null, subtree, null, null, true, null, null, null, null);
 758         Assert.assertEquals("Expecting one entity when searching for any TextData", 1, pager.getCount().intValue());
 759         SearchResult<TaxonBase> searchResult = pager.getRecords().get(0);
 760         Assert.assertTrue("the map of highlighted fragments should contain at least one item", searchResult.getFieldHighlightMap().size() > 0);
 761         String[] fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 762         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Abies</B>"));
 763
 764         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Pflanzenart Tannen", null, subtree, null, null, true, null, null, null, null);
 765         searchResult = pager.getRecords().get(0);
 766         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 767         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 768         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Pflanzenart</B>") || fragments[0].contains("<B>Tannen</B>"));
 769
 770         pager = taxonService.findByDescriptionElementFullText(TextData.class, "+Pflanzenart +Tannen", null, subtree, null, null, true, null, null, null, null);
 771         searchResult = pager.getRecords().get(0);
 772         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 773         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 774         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Pflanzenart</B>") && fragments[0].contains("<B>Tannen</B>"));
 775
 776         pager = taxonService.findByDescriptionElementFullText(TextData.class, "\"Pflanzenart aus der Gattung der Tannen\"", null, subtree, null, null, true, null, null, null, null);
 777         searchResult = pager.getRecords().get(0);
 778         Assert.assertTrue("Phrase search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 779         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 780         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>"));
 781
 782         pager = taxonService.findByDescriptionElementFullText(TextData.class, "Gatt*", null, subtree, null, null, true, null, null, null, null);
 783         searchResult = pager.getRecords().get(0);
 784         Assert.assertTrue("Wildcard search : Expecting at least one item in highlighted fragments", searchResult.getFieldHighlightMap().size() > 0);
 785         fragments = searchResult.getFieldHighlightMap().values().iterator().next();
 786         Assert.assertTrue("first fragments should contains serch term", fragments[0].contains("<B>Gatt"));
 787     }
 788
 789
 790     @Test
 791     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class)
 792     public final void testFindByFullText() throws IOException, LuceneParseException {
 793
 794         refreshLuceneIndex();
 795
 796         classificationService.find(CLASSIFICATION_UUID);
 797         TaxonNode subtree = null;
 798
 799         boolean NO_UNPUBLISHED = false;
 800
 801         Pager<SearchResult<TaxonBase>> pager = taxonService.findByFullText(null, "Abies", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 7
 802 //        logFreeTextSearchResults(pager, Level.DEBUG, null);
 803         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 804
 805
 806         pager = taxonService.findByFullText(null, "Abies", null, subtree, NO_UNPUBLISHED, null, true, null, null, null, null); // --> 7
 807 //        logFreeTextSearchResults(pager, Level.DEBUG, null);
 808         Assert.assertEquals("Expecting 6 entities", 6, pager.getCount().intValue());
 809         Synonym abiesSubalpina = (Synonym)taxonService.find(ABIES_SUBALPINA_UUID);
 810
 811         //accepted published, syn not published
 812         abiesSubalpina.getAcceptedTaxon().setPublish(true);
 813         commitAndStartNewTransaction();
 814         pager = taxonService.findByFullText(null, "Abies", null, subtree, NO_UNPUBLISHED, null, true, null, null, null, null); // --> 7
 815         Assert.assertEquals("Expecting 7 entities", 7, pager.getCount().intValue());
 816
 817         //accepted published, syn published
 818         abiesSubalpina = (Synonym)taxonService.find(abiesSubalpina.getUuid());
 819         abiesSubalpina.setPublish(true);
 820         commitAndStartNewTransaction();
 821         pager = taxonService.findByFullText(null, "Abies", null, subtree, NO_UNPUBLISHED, null, true, null, null, null, null); // --> 7
 822         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 823
 824         //accepted not published, syn published
 825         abiesSubalpina = (Synonym)taxonService.find(abiesSubalpina.getUuid());
 826         abiesSubalpina.getAcceptedTaxon().setPublish(false);
 827         commitAndStartNewTransaction();
 828         pager = taxonService.findByFullText(null, "Abies", null, subtree, NO_UNPUBLISHED, null, true, null, null, null, null); // --> 7
 829         Assert.assertEquals("Expecting 6 entities. Synonym and accepted should not be found, though synonym is published",
 830                 6, pager.getCount().intValue());
 831
 832         pager = taxonService.findByFullText(Taxon.class, "Abies", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 6
 833         Assert.assertEquals("Expecting 7 entities", 7, pager.getCount().intValue());
 834
 835         pager = taxonService.findByFullText(Synonym.class, "Abies", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 1
 836         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 837         pager = taxonService.findByFullText(Synonym.class, "Abies", null, subtree, NO_UNPUBLISHED, null, true, null, null, null, null); // --> 1
 838         Assert.assertEquals("Expecting 0 entity", 0, pager.getCount().intValue());
 839
 840         pager = taxonService.findByFullText(TaxonBase.class, "sec", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 7
 841         Assert.assertEquals("Expecting 8 entities", 9, pager.getCount().intValue());
 842
 843         pager = taxonService.findByFullText(null, "genus", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 1
 844         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 845
 846         pager = taxonService.findByFullText(Taxon.class, "subalpina", null, subtree, includeUnpublished, null, true, null, null, null, null); // --> 0
 847         Assert.assertEquals("Expecting 0 entities", 0, pager.getCount().intValue());
 848
 849         // synonym in classification ???
 850     }
 851
 852     @Test
 853     @DataSet
 854     public final void testPrepareByAreaSearch() throws IOException, LuceneParseException {
 855
 856         List<PresenceAbsenceTerm> statusFilter = new ArrayList<>();
 857         List<NamedArea> areaFilter = new ArrayList<>();
 858         areaFilter.add(germany);
 859         areaFilter.add(canada);
 860         areaFilter.add(russia);
 861
 862         Pager<SearchResult<TaxonBase>> pager = taxonService.findByDistribution(areaFilter, statusFilter, null, 20, 0, null, null);
 863         Assert.assertEquals("Expecting 2 entities", Integer.valueOf(2), Integer.valueOf(pager.getRecords().size()));
 864
 865     }
 866
 867     @Test
 868     @DataSet
 869     public final void testFindTaxaAndNamesByFullText() throws IOException, LuceneParseException, LuceneMultiSearchException {
 870
 871         refreshLuceneIndex();
 872         TaxonNode subtree = null;
 873
 874         Classification alternateClassification = classificationService.find(CLASSIFICATION_ALT_UUID);
 875         Synonym abiesSubalpina = (Synonym)taxonService.find(ABIES_SUBALPINA_UUID);
 876
 877         Pager<SearchResult<TaxonBase>> pager;
 878          pager = taxonService.findTaxaAndNamesByFullText(
 879                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
 880                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 881 //        logPagerRecords(pager, Level.DEBUG);
 882         Assert.assertEquals("doTaxa & doSynonyms & unpublished", 8, pager.getCount().intValue());
 883         pager = taxonService.findTaxaAndNamesByFullText(TaxaAndNamesSearchMode.taxaAndSynonyms(),
 884                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 885         Assert.assertEquals("doTaxa & doSynonyms, published only", 6, pager.getCount().intValue());
 886
 887         //accepted published, syn not published
 888         abiesSubalpina.getAcceptedTaxon().setPublish(true);
 889         commitAndStartNewTransaction();
 890         pager = taxonService.findTaxaAndNamesByFullText(TaxaAndNamesSearchMode.taxaAndSynonyms(),
 891                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 892         Assert.assertEquals("doTaxa & doSynonyms, accepted published", 7, pager.getCount().intValue());
 893
 894         //accepted published, syn published
 895         abiesSubalpina = (Synonym)taxonService.find(abiesSubalpina.getUuid());
 896         abiesSubalpina.setPublish(true);
 897         commitAndStartNewTransaction();
 898         pager = taxonService.findTaxaAndNamesByFullText(TaxaAndNamesSearchMode.taxaAndSynonyms(),
 899                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 900         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
 901
 902         //accepted not published, syn published
 903         abiesSubalpina = (Synonym)taxonService.find(abiesSubalpina.getUuid());
 904         abiesSubalpina.getAcceptedTaxon().setPublish(false);
 905         commitAndStartNewTransaction();
 906         pager = taxonService.findTaxaAndNamesByFullText(TaxaAndNamesSearchMode.taxaAndSynonyms(),
 907                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 908         Assert.assertEquals("Expecting 6 entities. Synonym and accepted should not be found, though synonym is published",
 909                 6, pager.getCount().intValue());
 910
 911         EnumSet<TaxaAndNamesSearchMode> searchMode = EnumSet.allOf(TaxaAndNamesSearchMode.class);
 912         pager = taxonService.findTaxaAndNamesByFullText(
 913                 searchMode, "Abies", null, subtree, null, null, null, true, null, null, null, null);
 914 //        logPagerRecords(pager, Level.DEBUG);
 915         Assert.assertEquals("all search modes", 8, pager.getCount().intValue());
 916         searchMode.remove(TaxaAndNamesSearchMode.includeUnpublished);
 917         pager = taxonService.findTaxaAndNamesByFullText(
 918                 searchMode, "Abies", null, subtree, null, null, null, true, null, null, null, null);
 919         Assert.assertEquals("all search modes except unpublished", 6, pager.getCount().intValue());
 920
 921         pager = taxonService.findTaxaAndNamesByFullText(EnumSet.allOf(TaxaAndNamesSearchMode.class),
 922                 "Abies", alternateClassification, subtree, null, null, null, true, null, null, null, null);
 923 //        logPagerRecords(pager, Level.DEBUG);
 924         Assert.assertEquals("all search modes, filtered by alternateClassification", 1, pager.getCount().intValue());
 925
 926         pager = taxonService.findTaxaAndNamesByFullText(
 927                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
 928                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 929         Assert.assertEquals("Expecting 2 entity", 2, pager.getCount().intValue());
 930         Set<UUID> uuids = getTaxonUuidSet(pager);
 931         Assert.assertTrue("The real synonym should be contained", uuids.contains(ABIES_SUBALPINA_UUID));
 932         Assert.assertTrue("The pro parte synonym should be contained",uuids.contains(ABIES_LASIOCARPA_UUID));
 933         //without published
 934         pager = taxonService.findTaxaAndNamesByFullText(EnumSet.of(TaxaAndNamesSearchMode.doSynonyms),
 935                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 936         Assert.assertEquals("Expecting 0 entities", 0, pager.getCount().intValue());
 937
 938
 939         pager = taxonService.findTaxaAndNamesByFullText(
 940                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames, TaxaAndNamesSearchMode.includeUnpublished),
 941                 "Abies", null, subtree, null, null, null, true, null, null, null, null);
 942         Assert.assertEquals("Expecting 0 entity", 0, pager.getCount().intValue());
 943
 944         pager = taxonService.findTaxaAndNamesByFullText(
 945                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames, TaxaAndNamesSearchMode.includeUnpublished),
 946                 "Tanne", null, subtree, null, null, null, true, null, null, null, null);
 947         Assert.assertEquals("Expecting 1 entity", 1, pager.getRecords().size());
 948         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 949         pager = taxonService.findTaxaAndNamesByFullText(
 950                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames),
 951                 "Tanne", null, subtree, null, null, null, true, null, null, null, null);
 952         Assert.assertEquals("Expecting 0 entity", 0, pager.getRecords().size());
 953
 954         //misapplied names
 955         pager = taxonService.findTaxaAndNamesByFullText(
 956                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.includeUnpublished),
 957                 "kawakamii", (Classification)null, subtree, null, null, null, true, null, null, null, null);
 958         logFreeTextSearchResults(pager, Level.DEBUG, null);
 959         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
 960         //unpublish accepted taxon
 961         pager = taxonService.findTaxaAndNamesByFullText(
 962                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
 963                 "kawakamii", (Classification)null, subtree, null, null, null, true, null, null, null, null);
 964         Assert.assertEquals("Expecting 0 entities", 0, pager.getCount().intValue());
 965         //published accepted taxon/misapplied name
 966         Taxon abiesBalsamea = (Taxon)taxonService.find(ABIES_BALSAMEA_UUID);
 967         abiesBalsamea.setPublish(true);
 968         commitAndStartNewTransaction();
 969         pager = taxonService.findTaxaAndNamesByFullText(
 970                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
 971                 "kawakamii", (Classification)null, subtree, null, null, null, true, null, null, null, null);
 972         Assert.assertEquals("Expecting 1 entities", 1, pager.getCount().intValue());
 973         //unpublished misapplied name
 974         Taxon misapplied = (Taxon)taxonService.find(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID);
 975         misapplied.setPublish(false);
 976         commitAndStartNewTransaction();
 977         pager = taxonService.findTaxaAndNamesByFullText(
 978                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
 979                 "kawakamii", (Classification)null, subtree, null, null, null, true, null, null, null, null);
 980         Assert.assertEquals("Expecting 0 entities", 0, pager.getCount().intValue());
 981
 982     }
 983
 984     @Test
 985     @DataSet
 986     //test for https://dev.e-taxonomy.eu/redmine/issues/7486
 987     public final void testFindTaxaAndNamesByFullText_synonymsAndMisapplied_7486() throws IOException, LuceneParseException, LuceneMultiSearchException {
 988
 989         refreshLuceneIndex();
 990         TaxonNode subtree = null;
 991
 992         //misapplied names
 993         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
 994                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.includeUnpublished),
 995                 "Abies", (Classification)null, subtree, null, null, null, true, null, null, null, null);
 996         logFreeTextSearchResults(pager, Level.DEBUG, null);
 997         Assert.assertEquals("Expecting 3 entity", 3, pager.getCount().intValue());
 998         Set<UUID> uuids = getTaxonUuidSet(pager);
 999         Assert.assertTrue("The real synonym should be contained", uuids.contains(ABIES_SUBALPINA_UUID));
1000         Assert.assertTrue("The pro parte synonym should be contained",uuids.contains(ABIES_LASIOCARPA_UUID));
1001         Assert.assertTrue("The misapplied name should be contained",uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1002     }
1003
1004     @Test
1005     @DataSet
1006     public final void testFindTaxaAndNamesByFullText_PhraseQuery() throws IOException, LuceneParseException, LuceneMultiSearchException {
1007
1008         refreshLuceneIndex();
1009         TaxonNode subtree = null;
1010
1011         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
1012                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1013                 "\"Abies alba\"", null, subtree, null, null, null, true, null, null, null, null);
1014 //        logPagerRecords(pager, Level.DEBUG);
1015         Assert.assertEquals("doTaxa & doSynonyms with simple phrase query", 1, pager.getCount().intValue());
1016
1017         pager = taxonService.findTaxaAndNamesByFullText(
1018                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1019                 "\"Abies al*\"", null, subtree, null, null, null, true, null, null, null, null);
1020 //        logPagerRecords(pager, Level.DEBUG);
1021         Assert.assertEquals("doTaxa & doSynonyms with complex phrase query", 1, pager.getCount().intValue());
1022
1023         pager = taxonService.findTaxaAndNamesByFullText(
1024                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1025                 "\"Abies*\"", null, subtree, null, null, null, true, null, null, null, null);
1026 //        logPagerRecords(pager, Level.DEBUG);
1027         Assert.assertEquals("doTaxa & doSynonyms with simple phrase query", 8, pager.getCount().intValue());
1028
1029     }
1030
1031     @Test
1032     @DataSet
1033     public final void testFindTaxaAndNamesByFullText_Sort() throws IOException, LuceneParseException, LuceneMultiSearchException {
1034
1035         refreshLuceneIndex();
1036         TaxonNode subtree = null;
1037
1038         List<OrderHint> orderHints = new ArrayList<>();
1039
1040 //        String[] docFields2log = new String[]{"id"};
1041
1042         // SortById
1043         orderHints.addAll(OrderHint.ORDER_BY_ID.asList());
1044         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
1045                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa),
1046                 "Abies", null, subtree, null, null, null, true, null, null, orderHints, null);
1047 //        logSearchResults(pager, Level.DEBUG, docFields2log);
1048         int 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         orderHints.addAll(OrderHint.ORDER_BY_ID.asList());
1057         pager = taxonService.findTaxaAndNamesByFullText(
1058                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1059                 "Abies", null, subtree, null, null, null, true, null, null, orderHints, null);
1060 //        logSearchResults(pager, Level.DEBUG, docFields2log);
1061
1062         lastId = -1;
1063         for(SearchResult<TaxonBase> rs : pager.getRecords()){
1064             if(lastId != -1){
1065                 Assert.assertTrue("results not sorted by id", lastId < rs.getEntity().getId());
1066             }
1067             lastId = rs.getEntity().getId();
1068         }
1069
1070         // Sortby NOMENCLATURAL_SORT_ORDER TODO make assertions !!!
1071         orderHints.clear();
1072         orderHints.addAll(OrderHint.NOMENCLATURAL_SORT_ORDER.asList());
1073         pager = taxonService.findTaxaAndNamesByFullText(
1074                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms),
1075                 "Abies", null, subtree, null, null, null, true, null, null, orderHints, null);
1076         logFreeTextSearchResults(pager, Level.DEBUG, null);
1077
1078     }
1079
1080     @Test
1081     @DataSet
1082     @Ignore //ignore until #7487 is fixed
1083     public final void testFindTaxaAndNamesByFullText_AreaFilter_7487() throws IOException, LuceneParseException, LuceneMultiSearchException {
1084         refreshLuceneIndex();
1085         TaxonNode subtree = null;
1086         Set<NamedArea> a_germany_canada_russia = new HashSet<>();
1087         a_germany_canada_russia.add(germany);
1088         a_germany_canada_russia.add(canada);
1089         a_germany_canada_russia.add(russia);
1090
1091         Set<PresenceAbsenceTerm> present_native = new HashSet<>();
1092         present_native.add(PresenceAbsenceTerm.PRESENT());
1093         present_native.add(PresenceAbsenceTerm.NATIVE());
1094
1095         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
1096                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1097                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1098         Assert.assertEquals("Synonyms with matching area filter", 2, pager.getCount().intValue());
1099         Set<UUID> uuids = this.getTaxonUuidSet(pager);
1100         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1101         Assert.assertTrue("Pro parte synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1102
1103         //pro parte syn => partial syn
1104         Taxon t_abies_balsamea = (Taxon)taxonService.find(ABIES_BALSAMEA_UUID);
1105         Set<TaxonRelationship> relsTo = t_abies_balsamea.getProParteAndPartialSynonymRelations();
1106         Assert.assertEquals(1, relsTo.size());
1107         TaxonRelationship taxonRelation = relsTo.iterator().next();
1108         taxonRelation.setType(TaxonRelationshipType.PARTIAL_SYNONYM_FOR());
1109         taxonService.saveOrUpdate(t_abies_balsamea);
1110         commitAndStartNewTransaction(null);
1111
1112         //should give same results as above
1113         pager = taxonService.findTaxaAndNamesByFullText(
1114                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1115                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1116 //        Assert.assertEquals("Synonyms with matching area filter", 2, pager.getCount().intValue());
1117 //        uuids = this.getTaxonUuidSet(pager);
1118 //        Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1119 //        Assert.assertTrue("Partial synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1120
1121         ///MISAPPLIED
1122         pager = taxonService.findTaxaAndNamesByFullText(
1123                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.includeUnpublished),
1124                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1125         Assert.assertEquals("misappliedNames with matching area & status filter", 3, pager.getCount().intValue());
1126         uuids = this.getTaxonUuidSet(pager);
1127         Assert.assertTrue("Misapplied name should be in", uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1128
1129         t_abies_balsamea = (Taxon)taxonService.find(ABIES_BALSAMEA_UUID);
1130         relsTo = t_abies_balsamea.getMisappliedNameRelations();
1131         Assert.assertEquals(1, relsTo.size());
1132         taxonRelation = relsTo.iterator().next();
1133         Assert.assertEquals(taxonRelation.getFromTaxon().getUuid(), D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID);
1134         taxonRelation.setType(TaxonRelationshipType.PRO_PARTE_MISAPPLIED_NAME_FOR());
1135         taxonService.saveOrUpdate(t_abies_balsamea);
1136         commitAndStartNewTransaction(null);
1137
1138         //strange it works here before fixing #7487 already
1139         pager = taxonService.findTaxaAndNamesByFullText(
1140                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.includeUnpublished),
1141                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1142         Assert.assertEquals("misappliedNames with matching area & status filter", 3, pager.getCount().intValue());
1143         uuids = this.getTaxonUuidSet(pager);
1144         Assert.assertTrue("Pro parte misapplied name should be in", uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1145
1146     }
1147
1148
1149     @Test
1150     @DataSet
1151     public final void testFindTaxaAndNamesByFullText_AreaFilter() throws IOException, LuceneParseException, LuceneMultiSearchException {
1152
1153         refreshLuceneIndex();
1154         TaxonNode subtree = null;
1155
1156         Set<NamedArea> a_germany_canada_russia = new HashSet<>();
1157         a_germany_canada_russia.add(germany);
1158         a_germany_canada_russia.add(canada);
1159         a_germany_canada_russia.add(russia);
1160
1161         Set<NamedArea> a_russia = new HashSet<>();
1162         a_russia.add(russia);
1163
1164         Set<PresenceAbsenceTerm> present = new HashSet<>();
1165         present.add(PresenceAbsenceTerm.PRESENT());
1166
1167         Set<PresenceAbsenceTerm> present_native = new HashSet<>();
1168         present_native.add(PresenceAbsenceTerm.PRESENT());
1169         present_native.add(PresenceAbsenceTerm.NATIVE());
1170
1171         Set<PresenceAbsenceTerm> absent = new HashSet<>();
1172         absent.add(PresenceAbsenceTerm.ABSENT());
1173
1174         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
1175                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.includeUnpublished),
1176                 "Abies", null, subtree, a_germany_canada_russia, null, null, true, null, null, null, null);
1177         logFreeTextSearchResults(pager, Level.DEBUG, null);
1178         Assert.assertEquals("Synonyms with matching area filter", 2, pager.getCount().intValue());
1179         Set<UUID> uuids = this.getTaxonUuidSet(pager);
1180         Assert.assertTrue("Abies alba (accepted with distribution) should be in", uuids.contains(ABIES_ALBA_UUID));
1181         Assert.assertTrue("Abies balsamea (accepted with distribution) should be in", uuids.contains(ABIES_BALSAMEA_UUID));
1182
1183         pager = taxonService.findTaxaAndNamesByFullText(
1184                 EnumSet.of(TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1185                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1186         Assert.assertEquals("Synonyms with matching area filter", 2, pager.getCount().intValue());
1187         uuids = this.getTaxonUuidSet(pager);
1188         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1189         Assert.assertTrue("Pro parte synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1190
1191         pager = taxonService.findTaxaAndNamesByFullText(
1192                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1193                 "Abies", null, subtree, a_germany_canada_russia, null, null, true, null, null, null, null);
1194         logFreeTextSearchResults(pager, Level.DEBUG, null);
1195         Assert.assertEquals("taxa and synonyms with matching area filter", 4, pager.getCount().intValue());
1196         uuids = this.getTaxonUuidSet(pager);
1197         Assert.assertTrue("Accepted taxon with area should be in", uuids.contains(ABIES_ALBA_UUID));
1198         Assert.assertTrue("Accepted taxon with area should be in", uuids.contains(ABIES_BALSAMEA_UUID));
1199         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1200         Assert.assertTrue("Pro parte synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1201         Assert.assertFalse("Misapplied name should NOT be in", uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1202
1203         pager = taxonService.findTaxaAndNamesByFullText(
1204                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1205                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1206         Assert.assertEquals("taxa and synonyms with matching area & status filter 4", 4, pager.getCount().intValue());
1207         uuids = this.getTaxonUuidSet(pager);
1208         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1209         Assert.assertTrue("Accepted taxon with area should be in", uuids.contains(ABIES_ALBA_UUID));
1210         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1211         Assert.assertTrue("Pro parte synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1212
1213         pager = taxonService.findTaxaAndNamesByFullText(
1214                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1215                 "Abies", null, subtree, a_germany_canada_russia, present, null, true, null, null, null, null);
1216         Assert.assertEquals("taxa and synonyms with matching area & status filter 3", 3, pager.getCount().intValue());
1217         uuids = this.getTaxonUuidSet(pager);
1218         Assert.assertTrue("Abies balsamea (accepted taxon) should be in", uuids.contains(ABIES_BALSAMEA_UUID));
1219         Assert.assertTrue("Synonym of balsamea should be in", uuids.contains(ABIES_SUBALPINA_UUID));
1220         Assert.assertTrue("Pro parte synonym of balsamea should be in", uuids.contains(ABIES_LASIOCARPA_UUID));
1221
1222         pager = taxonService.findTaxaAndNamesByFullText(
1223                 EnumSet.of(TaxaAndNamesSearchMode.doTaxa, TaxaAndNamesSearchMode.doSynonyms, TaxaAndNamesSearchMode.includeUnpublished),
1224                 "Abies", null, subtree, a_russia, present, null, true, null, null, null, null);
1225         Assert.assertEquals("taxa and synonyms with non matching area & status filter", 0, pager.getCount().intValue());
1226
1227         pager = taxonService.findTaxaAndNamesByFullText(
1228                 EnumSet.of(TaxaAndNamesSearchMode.doTaxaByCommonNames, TaxaAndNamesSearchMode.includeUnpublished),
1229                 "Tanne", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1230         Assert.assertEquals("ByCommonNames with area filter", 1, pager.getCount().intValue());
1231         uuids = this.getTaxonUuidSet(pager);
1232         Assert.assertTrue("Abies balsamea should be in", uuids.contains(ABIES_BALSAMEA_UUID));
1233
1234         // abies_kawakamii_sensu_komarov as misapplied name for t_abies_balsamea
1235         pager = taxonService.findTaxaAndNamesByFullText(
1236                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.includeUnpublished),
1237                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1238         Assert.assertEquals("misappliedNames with matching area & status filter", 1, pager.getCount().intValue());
1239         uuids = this.getTaxonUuidSet(pager);
1240         Assert.assertTrue("Misapplied name should  be in", uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1241
1242
1243         // 1. remove existing taxon relation
1244         Taxon t_abies_balsamea = (Taxon)taxonService.find(ABIES_BALSAMEA_UUID);
1245         Set<TaxonRelationship> relsTo = t_abies_balsamea.getMisappliedNameRelations();
1246         Assert.assertEquals(1, relsTo.size());
1247         TaxonRelationship taxonRelation = relsTo.iterator().next();
1248         t_abies_balsamea.removeTaxonRelation(taxonRelation);
1249         taxonService.saveOrUpdate(t_abies_balsamea);
1250         commitAndStartNewTransaction(null);
1251
1252         pager = taxonService.findTaxaAndNamesByFullText(
1253                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.includeUnpublished),
1254                 "Abies", null, subtree, a_germany_canada_russia, present_native, null, true, null, null, null, null);
1255         Assert.assertEquals("misappliedNames with matching area & status filter, should match nothing now", 0, pager.getCount().intValue());
1256
1257         // 2. now add abies_kawakamii_sensu_komarov as misapplied name for t_abies_alba and search for misapplications in Russia: ABSENT
1258         Taxon t_abies_kawakamii_sensu_komarov = (Taxon)taxonService.find(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID);
1259         Taxon t_abies_alba = (Taxon)taxonService.find(ABIES_ALBA_UUID);
1260         t_abies_alba.addMisappliedName(t_abies_kawakamii_sensu_komarov, null, null);
1261         taxonService.update(t_abies_kawakamii_sensu_komarov);
1262         commitAndStartNewTransaction(null);
1263
1264         pager = taxonService.findTaxaAndNamesByFullText(
1265                 EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames, TaxaAndNamesSearchMode.includeUnpublished),
1266                 "Abies", null, subtree, a_germany_canada_russia, absent, null, true, null, null, null, null);
1267         Assert.assertEquals("misappliedNames with matching area & status filter, should find one", 1, pager.getCount().intValue());
1268         uuids = this.getTaxonUuidSet(pager);
1269         Assert.assertTrue("Misapplied name should  be in", uuids.contains(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID));
1270
1271     }
1272
1273     @Test
1274     @DataSet
1275     //http://dev.e-taxonomy.eu/trac/ticket/5477
1276     public final void testFindTaxaAndNamesByFullText_AreaFilter_issue5477() throws IOException, LuceneParseException, LuceneMultiSearchException {
1277
1278         TaxonNode subtree = null;
1279         Set<NamedArea> a_germany_canada_russia = new HashSet<>();
1280         a_germany_canada_russia.add(germany);
1281         a_germany_canada_russia.add(canada);
1282         a_germany_canada_russia.add(russia);
1283
1284
1285         Set<PresenceAbsenceTerm> absent = new HashSet<>();
1286         absent.add(PresenceAbsenceTerm.ABSENT());
1287
1288         Taxon t_abies_kawakamii_sensu_komarov = (Taxon)taxonService.find(D_ABIES_KAWAKAMII_SEC_KOMAROV_UUID);
1289         Taxon t_abies_alba = (Taxon)taxonService.find(ABIES_ALBA_UUID);
1290         t_abies_alba.addMisappliedName(t_abies_kawakamii_sensu_komarov, null, null);
1291
1292         /* Since the upgrade from hibernate search 4 to 5.5
1293          * triggering an update of t_abies_alba is no longer sufficient to also update the
1294          * document of t_abies_kawakamii_sensu_komarov in the lucene index.
1295          * the last test in testFindTaxaAndNamesByFullText_AreaFilter() failed in this case.
1296          * This situation is reproduced here:
1297          */
1298         taxonService.update(t_abies_alba);
1299
1300         commitAndStartNewTransaction(null);
1301
1302         Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(
1303                   EnumSet.of(TaxaAndNamesSearchMode.doMisappliedNames),
1304                   "Abies", null, subtree, a_germany_canada_russia, absent, null, true, null, null, null, null);
1305         Assert.assertEquals("misappliedNames with matching area & status filter, should find one", 1, pager.getCount().intValue());
1306     }
1307
1308
1309     /**
1310      * Regression test for #3119: fulltext search: Entity always null whatever search
1311      *
1312      * @throws IOException
1313      * @throws LuceneParseException
1314      * @throws LuceneMultiSearchException
1315      */
1316     @Test
1317     @DataSet
1318     public final void testFindByEverythingFullText() throws IOException, LuceneParseException, LuceneMultiSearchException {
1319
1320         refreshLuceneIndex();
1321         TaxonNode subtree = null;
1322         EnumSet<TaxaAndNamesSearchMode> mode = TaxaAndNamesSearchMode.taxaAndSynonymsWithUnpublished();
1323         // via Taxon
1324         Pager<SearchResult<TaxonBase>>pager = taxonService.findByEverythingFullText("Abies", null, subtree, includeUnpublished, null, true, null, null, null, null);
1325 //        Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(mode,
1326 //                "Abies", null, null, null, null, true, null, null, null, null);
1327         logFreeTextSearchResults(pager, Level.DEBUG, null);
1328         Assert.assertTrue("Expecting at least 7 entities for 'Abies'", pager.getCount() > 7);
1329         Assert.assertNotNull("Expecting entity", pager.getRecords().get(0).getEntity());
1330 //        Assert.assertEquals("Expecting Taxon entity", Taxon.class, pager.getRecords().get(0).getEntity().getClass());
1331
1332         // via DescriptionElement
1333         pager = taxonService.findByEverythingFullText("present", null, subtree, includeUnpublished, null, true, null, null, null, null);
1334         //this is not covered by findTaxaAndNamesByFullText
1335 //        pager = taxonService.findTaxaAndNamesByFullText(mode,
1336 //                "present", null, null, null, null, true, null, null, null, null);
1337         Assert.assertEquals("Expecting one entity when searching for area 'present'", 1, pager.getCount().intValue());
1338         Assert.assertNotNull("Expecting entity", pager.getRecords().get(0).getEntity());
1339         Assert.assertEquals("Expecting Taxon entity", Taxon.class, CdmBase.deproxy(pager.getRecords().get(0).getEntity()).getClass());
1340         Assert.assertEquals("Expecting Taxon ", ABIES_BALSAMEA_UUID, pager.getRecords().get(0).getEntity().getUuid());
1341
1342     }
1343
1344
1345     @Test
1346     @DataSet
1347     public final void findByEveryThingFullText() throws IOException, LuceneParseException, LuceneMultiSearchException {
1348
1349         refreshLuceneIndex();
1350         TaxonNode subtree = null;
1351
1352         Classification classification = null;
1353         EnumSet<TaxaAndNamesSearchMode> mode = TaxaAndNamesSearchMode.taxaAndSynonymsWithUnpublished();
1354
1355         Pager<SearchResult<TaxonBase>> pager = taxonService.findByEverythingFullText("genus", null, subtree, includeUnpublished, null, false, null, null, null, null); // --> 1
1356 //        Pager<SearchResult<TaxonBase>> pager = taxonService.findTaxaAndNamesByFullText(mode,
1357 //                "genus", classification, null, null, null, false, null, null, null, null);
1358         Assert.assertEquals("Expecting 1 entity", 1, pager.getCount().intValue());
1359
1360         //FIXME FAILS: abies balamea is returned twice, see also testFullText_Grouping()
1361         pager = taxonService.findByEverythingFullText("Balsam", null, subtree, includeUnpublished, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
1362         logFreeTextSearchResults(pager, Level.DEBUG, null);
1363 //        pager = taxonService.findTaxaAndNamesByFullText(EnumSet.allOf(TaxaAndNamesSearchMode.class),
1364 //                "Balsam", classification, null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
1365         Assert.assertEquals("expecting to find the Abies balsamea via the GERMAN DescriptionElements", 1, pager.getCount().intValue());
1366
1367         //TODO fieldHighlight does not yet work
1368         pager = taxonService.findByEverythingFullText("Abies", null, subtree, includeUnpublished, null, true, null, null, null, null);
1369 //        pager = taxonService.findTaxaAndNamesByFullText(mode,
1370 //                "Abies", classification, null, null, Arrays.asList(new Language[]{Language.GERMAN()}), false, null, null, null, null);
1371         Assert.assertEquals("Expecting 8 entities", 8, pager.getCount().intValue());
1372         SearchResult<TaxonBase> searchResult = pager.getRecords().get(0);
1373         Assert.assertTrue("the map of highlighted fragments should contain at least one item", searchResult.getFieldHighlightMap().size() > 0);
1374         String[] fragments = searchResult.getFieldHighlightMap().values().iterator().next();
1375         Assert.assertTrue("first fragments should contains serch term", fragments[0].toLowerCase().contains("<b>abies</b>"));
1376     }
1377
1378 //    @SuppressWarnings("rawtypes")
1379 //    @Test
1380 //    @DataSet
1381 //    public final void benchmarkFindTaxaAndNamesHql() throws IOException, LuceneParseException {
1382 //
1383 //        createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1384 //
1385 //        IFindTaxaAndNamesConfigurator configurator = new FindTaxaAndNamesConfiguratorImpl();
1386 //        configurator.setTitleSearchString("Wei"+UTF8.SHARP_S+"%");
1387 //        configurator.setMatchMode(MatchMode.BEGINNING);
1388 //        configurator.setDoTaxa(false);
1389 //        configurator.setDoSynonyms(false);
1390 //        configurator.setDoNamesWithoutTaxa(false);
1391 //        configurator.setDoTaxaByCommonNames(true);
1392 //
1393 //        Pager<IdentifiableEntity> pager;
1394 //
1395 //        long startMillis = System.currentTimeMillis();
1396 //        for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1397 //            pager = taxonService.findTaxaAndNames(configurator);
1398 //            if (logger.isDebugEnabled()) {
1399 //                logger.debug("[" + indx + "]" + pager.getRecords().get(0).getTitleCache());
1400 //            }
1401 //        }
1402 //        double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1403 //        logger.info("Benchmark result - [find taxon by CommonName via HQL] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1404 //    }
1405
1406     @SuppressWarnings("rawtypes")
1407     @Test
1408     @DataSet
1409     public final void benchmarkFindByCommonNameHql() {
1410
1411 //        printDataSet(System.err, new String[] { "TaxonBase" });
1412
1413         createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1414
1415         IFindTaxaAndNamesConfigurator configurator = FindTaxaAndNamesConfiguratorImpl.NewInstance();
1416         configurator.setTitleSearchString("Wei"+UTF8.SHARP_S+"%");
1417         configurator.setMatchMode(MatchMode.BEGINNING);
1418         configurator.setDoTaxa(false);
1419         configurator.setDoSynonyms(false);
1420         configurator.setDoNamesWithoutTaxa(false);
1421         configurator.setDoTaxaByCommonNames(true);
1422
1423         Pager<IdentifiableEntity> pager;
1424
1425         long startMillis = System.currentTimeMillis();
1426         for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1427             pager = taxonService.findTaxaAndNames(configurator);
1428             if (logger.isDebugEnabled()) {
1429                 logger.debug("[" + indx + "]" + pager.getRecords().get(0).getTitleCache());
1430             }
1431         }
1432         double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1433         logger.info("Benchmark result - [find taxon by CommonName via HQL] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1434     }
1435
1436     @SuppressWarnings("rawtypes")
1437     @Test
1438     @DataSet
1439     public final void benchmarkFindByCommonNameLucene() throws IOException, LuceneParseException {
1440         TaxonNode subtree = null;
1441         createRandomTaxonWithCommonName(NUM_OF_NEW_RADOM_ENTITIES);
1442
1443         refreshLuceneIndex();
1444
1445         Pager<SearchResult<TaxonBase>> pager;
1446
1447         long startMillis = System.currentTimeMillis();
1448         for (int indx = 0; indx < BENCHMARK_ROUNDS; indx++) {
1449             pager = taxonService.findByDescriptionElementFullText(CommonTaxonName.class, "Wei"+UTF8.SHARP_S+"*", null, subtree, null, null, false, null, null, null, null);
1450             if (logger.isDebugEnabled()) {
1451                 logger.debug("[" + indx + "]" + pager.getRecords().get(0).getEntity().getTitleCache());
1452             }
1453         }
1454         double duration = ((double) (System.currentTimeMillis() - startMillis)) / BENCHMARK_ROUNDS;
1455         logger.info("Benchmark result - [find taxon by CommonName via lucene] : " + duration + "ms (" + BENCHMARK_ROUNDS + " benchmark rounds )");
1456     }
1457
1458     /**
1459      * uncomment @Test annotation to create the dataset for this test
1460      */
1461     @Override
1462     //    @Test
1463     @DataSet(loadStrategy=CleanSweepInsertLoadStrategy.class, value="BlankDataSet.xml")
1464     public final void createTestDataSet() throws FileNotFoundException {
1465
1466         Classification europeanAbiesClassification = Classification.NewInstance("European Abies");
1467         europeanAbiesClassification.setUuid(CLASSIFICATION_UUID);
1468         classificationService.save(europeanAbiesClassification);
1469
1470         Classification alternativeClassification = Classification.NewInstance("Abies alternative");
1471         alternativeClassification.setUuid(CLASSIFICATION_ALT_UUID);
1472         classificationService.save(alternativeClassification);
1473
1474         Reference sec = ReferenceFactory.newBook();
1475         sec.setTitleCache("Kohlbecker, A., Testcase standart views, 2013", true);
1476         Reference sec_sensu = ReferenceFactory.newBook();
1477         sec_sensu.setTitleCache("Komarov, V. L., Flora SSSR 29", true);
1478         referenceService.save(sec);
1479         referenceService.save(sec_sensu);
1480
1481         IBotanicalName n_abies = TaxonNameFactory.NewBotanicalInstance(Rank.GENUS());
1482         n_abies.setNameCache("Abies", true);
1483         Taxon t_abies = Taxon.NewInstance(n_abies, sec);
1484         taxonService.save(t_abies);
1485
1486         IBotanicalName n_abies_alba = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1487         n_abies_alba.setNameCache("Abies alba", true);
1488         Taxon t_abies_alba = Taxon.NewInstance(n_abies_alba, sec);
1489         t_abies_alba.setUuid(ABIES_ALBA_UUID);
1490         taxonService.save(t_abies_alba);
1491
1492         IBotanicalName n_abies_subalpina = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1493         n_abies_subalpina.setNameCache("Abies subalpina", true);
1494         Synonym s_abies_subalpina = Synonym.NewInstance(n_abies_subalpina, sec);
1495         taxonService.save(s_abies_subalpina);
1496
1497         IBotanicalName n_abies_balsamea = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1498         n_abies_balsamea.setNameCache("Abies balsamea", true);
1499         Taxon t_abies_balsamea = Taxon.NewInstance(n_abies_balsamea, sec);
1500         t_abies_balsamea.setUuid(ABIES_BALSAMEA_UUID);
1501         t_abies_balsamea.addSynonym(s_abies_subalpina, SynonymType.SYNONYM_OF());
1502         taxonService.save(t_abies_balsamea);
1503
1504         IBotanicalName n_abies_grandis = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1505         n_abies_grandis.setNameCache("Abies grandis", true);
1506         Taxon t_abies_grandis = Taxon.NewInstance(n_abies_grandis, sec);
1507         taxonService.save(t_abies_grandis);
1508
1509         IBotanicalName n_abies_kawakamii = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1510         n_abies_kawakamii.setNameCache("Abies kawakamii", true);
1511         Taxon t_abies_kawakamii = Taxon.NewInstance(n_abies_kawakamii, sec);
1512         t_abies_kawakamii.getTitleCache();
1513         taxonService.save(t_abies_kawakamii);
1514
1515         // abies_kawakamii_sensu_komarov as missapplied name for t_abies_balsamea
1516         Taxon t_abies_kawakamii_sensu_komarov = Taxon.NewInstance(n_abies_kawakamii, sec_sensu);
1517         taxonService.save(t_abies_kawakamii_sensu_komarov);
1518         t_abies_kawakamii_sensu_komarov.addTaxonRelation(t_abies_balsamea, TaxonRelationshipType.MISAPPLIED_NAME_FOR(), null, null);
1519         taxonService.saveOrUpdate(t_abies_kawakamii_sensu_komarov);
1520
1521         IBotanicalName n_abies_lasiocarpa = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1522         n_abies_lasiocarpa.setNameCache("Abies lasiocarpa", true);
1523         Taxon t_abies_lasiocarpa = Taxon.NewInstance(n_abies_lasiocarpa, sec);
1524         taxonService.save(t_abies_lasiocarpa);
1525
1526         // add taxa to classifications
1527         europeanAbiesClassification.addChildTaxon(t_abies_balsamea, null, null);
1528         alternativeClassification.addChildTaxon(t_abies_lasiocarpa, null, null);
1529         classificationService.saveOrUpdate(europeanAbiesClassification);
1530         classificationService.saveOrUpdate(alternativeClassification);
1531
1532         //
1533         // Description
1534         //
1535         TaxonDescription d_abies_alba = TaxonDescription.NewInstance(t_abies_alba);
1536         TaxonDescription d_abies_balsamea = TaxonDescription.NewInstance(t_abies_balsamea);
1537
1538         d_abies_alba.setUuid(D_ABIES_ALBA_UUID);
1539         d_abies_balsamea.setUuid(D_ABIES_BALSAMEA_UUID);
1540
1541
1542         // CommonTaxonName
1543         d_abies_alba.addElement(CommonTaxonName.NewInstance("Wei"+UTF8.SHARP_S+"tanne", Language.GERMAN()));
1544         d_abies_alba.addElement(CommonTaxonName.NewInstance("silver fir", Language.ENGLISH()));
1545         d_abies_alba.addElement(Distribution
1546                 .NewInstance(
1547                         germany,
1548                         PresenceAbsenceTerm.NATIVE()));
1549         d_abies_alba.addElement(Distribution
1550                 .NewInstance(
1551                         russia,
1552                         PresenceAbsenceTerm.ABSENT()));
1553
1554         // TextData
1555         d_abies_balsamea
1556             .addElement(TextData
1557                     .NewInstance(
1558                             "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.",
1559                             Language.GERMAN(), null));
1560         d_abies_balsamea
1561         .addElement(CommonTaxonName
1562                 .NewInstance(
1563                         "Balsam-Tanne",
1564                         Language.GERMAN(), null));
1565
1566         d_abies_balsamea
1567         .addElement(Distribution
1568                 .NewInstance(
1569                         canada,
1570                         PresenceAbsenceTerm.PRESENT()));
1571
1572         d_abies_balsamea
1573         .addElement(Distribution
1574                 .NewInstance(
1575                         germany,
1576                         PresenceAbsenceTerm.NATIVE()));
1577
1578         d_abies_balsamea
1579                 .addElement(TextData
1580                         .NewInstance(
1581                                 TaxonServiceSearchTestUtf8Constants.RUSSIAN_ABIES_ALBA_LONG,
1582                                 Language.RUSSIAN(), null));
1583         d_abies_balsamea
1584         .addElement(CommonTaxonName
1585                 .NewInstance(
1586                         TaxonServiceSearchTestUtf8Constants.RUSSIAN_ABIES_ALBA_SHORT,
1587                         Language.RUSSIAN(), null));
1588         descriptionService.saveOrUpdate(d_abies_balsamea);
1589
1590         setComplete();
1591         endTransaction();
1592
1593
1594         writeDbUnitDataSetFile(new String[] {
1595             "TAXONBASE", "TAXONNAME",
1596             "TAXONRELATIONSHIP",
1597             "REFERENCE", "DESCRIPTIONELEMENTBASE", "DESCRIPTIONBASE",
1598             "AGENTBASE", "HOMOTYPICALGROUP",
1599             "CLASSIFICATION", "TAXONNODE",
1600             "LANGUAGESTRING", "DESCRIPTIONELEMENTBASE_LANGUAGESTRING",
1601             "HIBERNATE_SEQUENCES" // IMPORTANT!!!
1602             });
1603
1604     }
1605
1606     /**
1607      *
1608      */
1609     private void refreshLuceneIndex() {
1610
1611 //        commitAndStartNewTransaction(null);
1612         commit();
1613         endTransaction();
1614         indexer.purge(DefaultProgressMonitor.NewInstance());
1615         indexer.reindex(typesToIndex, DefaultProgressMonitor.NewInstance());
1616         startNewTransaction();
1617 //        commitAndStartNewTransaction(null);
1618     }
1619
1620     /**
1621      * @param numberOfNew
1622      *
1623      */
1624     private void createRandomTaxonWithCommonName(int numberOfNew) {
1625
1626         logger.debug(String.format("creating %1$s random taxan with CommonName", numberOfNew));
1627
1628         commitAndStartNewTransaction(null);
1629
1630         Reference sec = ReferenceFactory.newBook();
1631         referenceService.save(sec);
1632
1633         for (int i = numberOfNew; i < numberOfNew; i++) {
1634             RandomStringUtils.randomAlphabetic(10);
1635             String radomName = RandomStringUtils.randomAlphabetic(5) + " " + RandomStringUtils.randomAlphabetic(10);
1636             String radomCommonName = RandomStringUtils.randomAlphabetic(10);
1637
1638             IBotanicalName name = TaxonNameFactory.NewBotanicalInstance(Rank.SPECIES());
1639             name.setNameCache(radomName, true);
1640             Taxon taxon = Taxon.NewInstance(name, sec);
1641             taxonService.save(taxon);
1642
1643             TaxonDescription description = TaxonDescription.NewInstance(taxon);
1644             description.addElement(CommonTaxonName.NewInstance(radomCommonName, Language.GERMAN()));
1645             descriptionService.save(description);
1646         }
1647
1648         commitAndStartNewTransaction(null);
1649     }
1650
1651     private <T extends CdmBase> void logFreeTextSearchResults(Pager<SearchResult<T>> pager, Level level, String[] docFields){
1652         if(level == null){
1653             level = Level.DEBUG;
1654         }
1655         if(logger.isEnabledFor(level)){
1656             StringBuilder b = new StringBuilder();
1657             b.append("\n");
1658             int i = 0;
1659             for(SearchResult<?> sr : pager.getRecords()){
1660
1661                 b.append(" ").append(i++).append(" - ");
1662                 b.append("score:").append(sr.getScore()).append(", ");
1663
1664                 if(docFields != null){
1665                     b.append("docs : ");
1666                     for(Document doc : sr.getDocs()) {
1667                         b.append("<");
1668                         for(String f : docFields){
1669                             b.append(f).append(":").append(Arrays.toString(doc.getValues(f)));
1670                         }
1671                         b.append(">");
1672                     }
1673                 }
1674
1675                 CdmBase entity = sr.getEntity();
1676                 if(entity == null){
1677                     b.append("NULL");
1678                 } else {
1679                     b.append(entity.getClass().getSimpleName()).
1680                         append(" [").append(entity.getId()).
1681                         append(" | ").append(entity.getUuid()).append("] : ").
1682                         append(entity.toString());
1683
1684                 }
1685                 b.append("\n");
1686             }
1687             logger.log(level, b);
1688         }
1689     }
1690
1691
1692     /**
1693      * @param pager
1694      * @return
1695      */
1696     private Set<UUID> getTaxonUuidSet(@SuppressWarnings("rawtypes") Pager<SearchResult<TaxonBase>> pager) {
1697         Set<UUID> result = new HashSet<>();
1698         for (@SuppressWarnings("rawtypes") SearchResult<TaxonBase> searchResult : pager.getRecords()){
1699             result.add(searchResult.getEntity().getUuid());
1700         }
1701         return result;
1702     }
1703
1704 }