cdm-eflora/src/main/java/eu/etaxonomy/cdm/io/eflora/centralAfrica/checklist/CentralAfricaChecklistTaxonImport.java

   1 /**
   2 * Copyright (C) 2007 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.io.eflora.centralAfrica.checklist;
  11
  12 import java.sql.ResultSet;
  13 import java.sql.SQLException;
  14 import java.util.HashMap;
  15 import java.util.HashSet;
  16 import java.util.Map;
  17 import java.util.Set;
  18 import java.util.UUID;
  19
  20 import org.apache.commons.lang.StringUtils;
  21 import org.apache.log4j.Logger;
  22 import org.springframework.stereotype.Component;
  23
  24 import eu.etaxonomy.cdm.api.service.IClassificationService;
  25 import eu.etaxonomy.cdm.common.CdmUtils;
  26 import eu.etaxonomy.cdm.io.common.IOValidator;
  27 import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
  28 import eu.etaxonomy.cdm.io.common.mapping.DbImportMapping;
  29 import eu.etaxonomy.cdm.io.common.mapping.DbImportMarkerMapper;
  30 import eu.etaxonomy.cdm.io.common.mapping.DbImportObjectCreationMapper;
  31 import eu.etaxonomy.cdm.io.common.mapping.DbImportTaxIncludedInMapper;
  32 import eu.etaxonomy.cdm.io.common.mapping.IMappingImport;
  33 import eu.etaxonomy.cdm.io.eflora.centralAfrica.checklist.validation.CentralAfricaChecklistTaxonImportValidator;
  34 import eu.etaxonomy.cdm.model.common.CdmBase;
  35 import eu.etaxonomy.cdm.model.description.Distribution;
  36 import eu.etaxonomy.cdm.model.description.PresenceTerm;
  37 import eu.etaxonomy.cdm.model.description.TaxonDescription;
  38 import eu.etaxonomy.cdm.model.location.NamedArea;
  39 import eu.etaxonomy.cdm.model.location.TdwgArea;
  40 import eu.etaxonomy.cdm.model.name.BotanicalName;
  41 import eu.etaxonomy.cdm.model.name.Rank;
  42 import eu.etaxonomy.cdm.model.reference.Reference;
  43 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  44 import eu.etaxonomy.cdm.model.taxon.Classification;
  45 import eu.etaxonomy.cdm.model.taxon.Taxon;
  46 import eu.etaxonomy.cdm.model.taxon.TaxonBase;
  47 import eu.etaxonomy.cdm.model.taxon.TaxonNode;
  48 import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
  49
  50
  51 /**
  52  * @author a.mueller
  53  * @created 20.02.2010
  54  * @version 1.0
  55  */
  56 @Component
  57 public class CentralAfricaChecklistTaxonImport  extends CentralAfricaChecklistImportBase<TaxonBase> implements IMappingImport<TaxonBase, CentralAfricaChecklistImportState>{
  58         private static final Logger logger = Logger.getLogger(CentralAfricaChecklistTaxonImport.class);
  59
  60         private NonViralNameParserImpl parser = NonViralNameParserImpl.NewInstance();
  61
  62         private Map<UUID, Taxon> higherTaxonMap;
  63
  64         private Integer TREE_ID = null;
  65
  66         private DbImportMapping mapping;
  67
  68         private int modCount = 10000;
  69         private static final String pluralString = "taxa";
  70         private static final String dbTableName = "checklist";
  71         private static final Class cdmTargetClass = TaxonBase.class;
  72         private static final String strOrderBy = " ORDER BY family, genus, species ";
  73
  74         public CentralAfricaChecklistTaxonImport(){
  75                 super(pluralString, dbTableName, cdmTargetClass);
  76         }
  77
  78
  79
  80         /* (non-Javadoc)
  81          * @see eu.etaxonomy.cdm.io.erms.ErmsImportBase#getIdQuery()
  82          */
  83         @Override
  84         protected String getIdQuery() {
  85                 String strQuery = " SELECT pk FROM " + dbTableName +
  86                                                 strOrderBy;
  87                 return strQuery;
  88         }
  89
  90
  91         /* (non-Javadoc)
  92          * @see eu.etaxonomy.cdm.io.erms.ErmsImportBase#getMapping()
  93          */
  94         protected DbImportMapping getMapping() {
  95                 if (mapping == null){
  96                         mapping = new DbImportMapping();
  97
  98                         mapping.addMapper(DbImportObjectCreationMapper.NewInstance(this, "pk", TAXON_NAMESPACE)); //id + tu_status
  99
 100                         UUID uuidKew = CentralAfricaChecklistTransformer.uuidAcceptedKew;
 101                         mapping.addMapper(DbImportMarkerMapper.NewInstance("accepted kew", uuidKew, "Accepted Kew", "Accepted Kew", "Kew", null));
 102
 103                         UUID uuidGeneva = CentralAfricaChecklistTransformer.uuidAcceptedGeneva;
 104                         mapping.addMapper(DbImportMarkerMapper.NewInstance("accepted geneva", uuidGeneva, "Accepted Geneva", "Accepted Geneva", "Geneva", null));
 105
 106                         UUID uuidItis = CentralAfricaChecklistTransformer.uuidAcceptedItis;
 107                         mapping.addMapper(DbImportMarkerMapper.NewInstance("accepted itis", uuidItis, "Accepted ITIS", "Accepted ITIS", "ITIS", null));
 108                 }
 109
 110                 return mapping;
 111         }
 112
 113         /* (non-Javadoc)
 114          * @see eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportBase#getRecordQuery(eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator)
 115          */
 116         @Override
 117         protected String getRecordQuery(CentralAfricaChecklistImportConfigurator config) {
 118                 String strSelect = " SELECT * ";
 119                 String strFrom = " FROM checklist";
 120                 String strWhere = " WHERE ( pk IN (" + ID_LIST_TOKEN + ") )";
 121                 String strRecordQuery = strSelect + strFrom + strWhere + strOrderBy;
 122                 return strRecordQuery;
 123         }
 124
 125
 126         @Override
 127         public boolean doPartition(ResultSetPartitioner partitioner, CentralAfricaChecklistImportState state) {
 128                 higherTaxonMap = new HashMap<UUID, Taxon>();
 129                 Reference genevaReference = getReferenceService().find(state.getConfig().getUuidGenevaReference());
 130                 if (genevaReference == null){
 131                         genevaReference = makeGenevaReference(state);
 132                         getReferenceService().save(genevaReference);
 133                 }
 134                 state.setGenevaReference(genevaReference);
 135                 boolean success = super.doPartition(partitioner, state);
 136                 higherTaxonMap = new HashMap<UUID, Taxon>();
 137                 state.setGenevaReference(null);
 138                 return success;
 139         }
 140
 141
 142         /* (non-Javadoc)
 143          * @see eu.etaxonomy.cdm.io.berlinModel.in.IPartitionedIO#getRelatedObjectsForPartition(java.sql.ResultSet)
 144          */
 145         public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs) {
 146                 String nameSpace;
 147                 Class cdmClass;
 148                 Set<String> idSet;
 149                 Set<String> referenceIdSet = new HashSet<String>();
 150
 151                 Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>();
 152
 153                 try{
 154                         while (rs.next()){
 155                                 handleForeignKey(rs, referenceIdSet, "source");
 156                         }
 157
 158                         //reference map
 159                         nameSpace = REFERENCE_NAMESPACE;
 160                         cdmClass = Reference.class;
 161                         idSet = referenceIdSet;
 162                         Map<String, Reference> referenceMap = (Map<String, Reference>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, referenceIdSet, nameSpace);
 163                         result.put(REFERENCE_NAMESPACE, referenceMap);
 164
 165                 } catch (SQLException e) {
 166                         throw new RuntimeException(e);
 167                 }
 168                 return result;
 169         }
 170
 171
 172         /* (non-Javadoc)
 173          * @see eu.etaxonomy.cdm.io.common.mapping.IMappingImport#createObject(java.sql.ResultSet)
 174          */
 175         public TaxonBase createObject(ResultSet rs, CentralAfricaChecklistImportState state) throws SQLException {
 176                 BotanicalName speciesName = BotanicalName.NewInstance(Rank.SPECIES());
 177
 178                 Reference sec = state.getConfig().getSourceReference();
 179                 getReferenceService().saveOrUpdate(sec);
 180
 181                 String familyString = rs.getString("family");
 182                 String genusString = rs.getString("genus");
 183                 String speciesString = rs.getString("species");
 184                 String authorityString = rs.getString("authority");
 185
 186                 if (logger.isDebugEnabled()){
 187                         System.out.println(familyString + " " + genusString + " " + speciesString);
 188                 }
 189
 190                 Taxon speciesTaxon = Taxon.NewInstance(speciesName, sec);;
 191                 speciesName.setGenusOrUninomial(genusString);
 192                 speciesName.setSpecificEpithet(speciesString);
 193                 parser.handleAuthors(speciesName, CdmUtils.concat(" ", new String[] {"", genusString, speciesString, authorityString}), authorityString);
 194
 195                 //family
 196                 Taxon familyTaxon = null;
 197                 if (StringUtils.isNotBlank(familyString)){
 198                         familyTaxon = getHigherTaxon(state, familyString, null);
 199                         if (familyTaxon == null){
 200                                 BotanicalName familyName = BotanicalName.NewInstance(Rank.FAMILY());
 201                                 familyName.setGenusOrUninomial(familyString);
 202                                 familyTaxon = Taxon.NewInstance(familyName, sec);
 203                                 saveHigherTaxon(state, familyTaxon, familyString, null);
 204                         }
 205                         getTaxonService().saveOrUpdate(familyTaxon);
 206                 }
 207
 208
 209                 //genus
 210                 Taxon genusTaxon = getHigherTaxon(state, familyString, genusString);
 211                 if (genusTaxon == null){
 212                         BotanicalName genusName = BotanicalName.NewInstance(Rank.GENUS());
 213                         genusName.setGenusOrUninomial(genusString);
 214                         genusTaxon = Taxon.NewInstance(genusName, sec);
 215                         saveHigherTaxon(state, genusTaxon, familyString, genusString);
 216                         if (familyTaxon != null){
 217                                 makeTaxonomicallyIncluded(state, TREE_ID, genusTaxon, familyTaxon, null, null);
 218                         }
 219                 }
 220                 makeTaxonomicallyIncluded(state, TREE_ID, speciesTaxon, genusTaxon, null, null);
 221                 getTaxonService().saveOrUpdate(genusTaxon);
 222
 223                 String sourceString = rs.getString("source");
 224                 String sourceId = rs.getString("source_id");
 225
 226                 Reference sourceRef = state.getRelatedObject(REFERENCE_NAMESPACE, sourceString, Reference.class);
 227                 speciesTaxon.addSource(sourceId, REFERENCE_NAMESPACE, sourceRef, null);
 228
 229
 230                 //geneva id
 231                 Reference genevaReference = state.getGenevaReference();
 232                 Object genevaId = rs.getObject("geneva_ID");
 233                 speciesTaxon.addSource(String.valueOf(genevaId), null, genevaReference, null);
 234
 235                 //distribution
 236                 handleDistribution(rs, speciesTaxon);
 237
 238                 return speciesTaxon;
 239         }
 240
 241         private void handleDistribution(ResultSet rs, Taxon speciesTaxon) throws SQLException {
 242                 TaxonDescription description = TaxonDescription.NewInstance(speciesTaxon);
 243
 244                 Boolean isCongo = rs.getBoolean("drc");
 245                 Boolean isBurundi = rs.getBoolean("burundi");
 246                 Boolean isRwanda = rs.getBoolean("rwanda");
 247
 248                 addDistribution(description, isCongo, "ZAI");
 249                 addDistribution(description, isBurundi, "BUR");
 250                 addDistribution(description, isRwanda, "RWA");
 251
 252         }
 253
 254
 255
 256         /**
 257          * @param description
 258          * @param isCongo
 259          */
 260         private void addDistribution(TaxonDescription description, Boolean exists, String label) {
 261                 if (exists == true){
 262                         NamedArea namedArea = TdwgArea.getAreaByTdwgAbbreviation(label);
 263                         Distribution distribution = Distribution.NewInstance(namedArea, PresenceTerm.PRESENT());
 264                         description.addElement(distribution);
 265                 }
 266         }
 267
 268
 269
 270         private void saveHigherTaxon(CentralAfricaChecklistImportState state, Taxon higherTaxon, String family, String genus) {
 271                 String higherName = normalizeHigherTaxonName(family, genus);
 272                 UUID uuid = higherTaxon.getUuid();
 273                 state.putHigherTaxon(higherName, uuid);
 274                 higherTaxonMap.put(uuid, higherTaxon);
 275         }
 276
 277
 278
 279         private Taxon getHigherTaxon(CentralAfricaChecklistImportState state, String family, String genus) {
 280                 String higherName = normalizeHigherTaxonName(family, genus);
 281                 UUID uuid = state.getHigherTaxon(higherName);
 282
 283                 Taxon taxon = null;
 284                 if (uuid != null){
 285                         taxon = higherTaxonMap.get(uuid);
 286                         if (taxon == null){
 287                                 taxon = CdmBase.deproxy(getTaxonService().find(uuid), Taxon.class);
 288                         }
 289                 }
 290                 return taxon;
 291         }
 292
 293
 294
 295         /**
 296          * @param family
 297          * @param genus
 298          */
 299         private String normalizeHigherTaxonName(String family, String genus) {
 300                 return (CdmUtils.Nz(family) + "-" + CdmUtils.Nz(genus)).trim();
 301         }
 302
 303
 304
 305
 306 //      private boolean makeTaxonomicallyIncluded(CentralAfricaChecklistImportState state, Taxon parent, Taxon child, Reference citation, String microCitation){
 307 //              Reference sec = child.getSec();
 308 //              UUID uuid = state.getTreeUuid(sec);
 309 //              Classification tree;
 310 //              tree = state.getTree(sec);
 311 //
 312 //              if (tree == null){
 313 //                      tree = makeTreeMemSave(state, sec);
 314 //              }
 315 //              TaxonNode childNode;
 316 //              if (parent != null){
 317 //                      childNode = tree.addParentChild(parent, child, citation, microCitation);
 318 //              }else{
 319 //                      childNode = tree.addChildTaxon(child, citation, microCitation, null);
 320 //              }
 321 //              return (childNode != null);
 322 //      }
 323
 324         //TODO use Mapper
 325         private boolean makeTaxonomicallyIncluded(CentralAfricaChecklistImportState state, Integer treeRefFk, Taxon child, Taxon parent, Reference citation, String microCitation){
 326                 String treeKey;
 327                 UUID treeUuid;
 328                 if (treeRefFk == null){
 329                         treeKey = "1";  // there is only one tree and it gets the map key '1'
 330                         treeUuid = state.getConfig().getClassificationUuid();
 331                 }else{
 332                         treeKey =String.valueOf(treeRefFk);
 333                         treeUuid = state.getTreeUuidByTreeKey(treeKey);
 334                 }
 335                 Classification tree = (Classification)state.getRelatedObject(DbImportTaxIncludedInMapper.TAXONOMIC_TREE_NAMESPACE, treeKey);
 336                 if (tree == null){
 337                         IClassificationService service = state.getCurrentIO().getClassificationService();
 338                         tree = service.find(treeUuid);
 339                         if (tree == null){
 340                                 String treeName = state.getConfig().getClassificationName();
 341                                 tree = Classification.NewInstance(treeName);
 342                                 tree.setUuid(treeUuid);
 343                                 //FIXME tree reference
 344                                 //tree.setReference(ref);
 345                                 service.save(tree);
 346                         }
 347                         state.addRelatedObject(DbImportTaxIncludedInMapper.TAXONOMIC_TREE_NAMESPACE, treeKey, tree);
 348                 }
 349
 350                 TaxonNode childNode = tree.addParentChild(parent, child, citation, microCitation);
 351                 return (childNode != null);
 352         }
 353
 354
 355         private Reference makeGenevaReference(CentralAfricaChecklistImportState state) {
 356                 Reference result = ReferenceFactory.newDatabase();
 357                 result.setTitleCache(state.getConfig().getGenevaReferenceTitle(), true);
 358                 result.setUuid(state.getConfig().getUuidGenevaReference());
 359                 return result;
 360         }
 361
 362         /* (non-Javadoc)
 363          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 364          */
 365         @Override
 366         protected boolean doCheck(CentralAfricaChecklistImportState state){
 367                 IOValidator<CentralAfricaChecklistImportState> validator = new CentralAfricaChecklistTaxonImportValidator();
 368                 return validator.validate(state);
 369         }
 370
 371
 372         /* (non-Javadoc)
 373          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 374          */
 375         protected boolean isIgnore(CentralAfricaChecklistImportState state){
 376                 return ! state.getConfig().isDoTaxa();
 377         }
 378
 379
 380
 381 }