app-import/src/main/java/eu/etaxonomy/cdm/io/globis/GlobisCommonNameImport.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.globis;
  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
  19 import org.apache.commons.lang.StringUtils;
  20 import org.apache.log4j.Logger;
  21 import org.springframework.stereotype.Component;
  22
  23 import eu.etaxonomy.cdm.common.CdmUtils;
  24 import eu.etaxonomy.cdm.io.common.IOValidator;
  25 import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
  26 import eu.etaxonomy.cdm.io.globis.validation.GlobisCurrentSpeciesImportValidator;
  27 import eu.etaxonomy.cdm.model.common.CdmBase;
  28 import eu.etaxonomy.cdm.model.common.Language;
  29 import eu.etaxonomy.cdm.model.description.Distribution;
  30 import eu.etaxonomy.cdm.model.description.PresenceTerm;
  31 import eu.etaxonomy.cdm.model.description.TaxonDescription;
  32 import eu.etaxonomy.cdm.model.location.WaterbodyOrCountry;
  33 import eu.etaxonomy.cdm.model.name.Rank;
  34 import eu.etaxonomy.cdm.model.name.ZoologicalName;
  35 import eu.etaxonomy.cdm.model.reference.Reference;
  36 import eu.etaxonomy.cdm.model.taxon.Classification;
  37 import eu.etaxonomy.cdm.model.taxon.Taxon;
  38 import eu.etaxonomy.cdm.model.taxon.TaxonBase;
  39 import eu.etaxonomy.cdm.model.taxon.TaxonNode;
  40
  41
  42 /**
  43  * @author a.mueller
  44  * @created 20.02.2010
  45  * @version 1.0
  46  */
  47 @Component
  48 public class GlobisCommonNameImport  extends GlobisImportBase<Taxon> {
  49         private static final Logger logger = Logger.getLogger(GlobisCommonNameImport.class);
  50
  51         private int modCount = 10000;
  52         private static final String pluralString = "common names";
  53         private static final String dbTableName = "species_language";
  54         private static final Class cdmTargetClass = Taxon.class;  //not needed
  55
  56         public GlobisCommonNameImport(){
  57                 super(pluralString, dbTableName, cdmTargetClass);
  58         }
  59
  60
  61
  62
  63         /* (non-Javadoc)
  64          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#getIdQuery()
  65          */
  66         @Override
  67         protected String getIdQuery() {
  68                 String strRecordQuery =
  69                         " SELECT ID " +
  70                         " FROM " + dbTableName;
  71                 return strRecordQuery;
  72         }
  73
  74
  75
  76
  77         /* (non-Javadoc)
  78          * @see eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportBase#getRecordQuery(eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator)
  79          */
  80         @Override
  81         protected String getRecordQuery(GlobisImportConfigurator config) {
  82                 String strRecordQuery =
  83                         " SELECT * " +
  84                         " FROM " + getTableName() + " sl " +
  85                         " WHERE ( sl.ID IN (" + ID_LIST_TOKEN + ") )";
  86                 return strRecordQuery;
  87         }
  88
  89
  90
  91         /* (non-Javadoc)
  92          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doPartition(eu.etaxonomy.cdm.io.common.ResultSetPartitioner, eu.etaxonomy.cdm.io.globis.GlobisImportState)
  93          */
  94         @Override
  95         public boolean doPartition(ResultSetPartitioner partitioner, GlobisImportState state) {
  96                 boolean success = true;
  97
  98                 Set<TaxonBase> objectsToSave = new HashSet<TaxonBase>();
  99
 100                 Map<String, Taxon> taxonMap = (Map<String, Taxon>) partitioner.getObjectMap(TAXON_NAMESPACE);
 101 //              Map<String, DerivedUnit> ecoFactDerivedUnitMap = (Map<String, DerivedUnit>) partitioner.getObjectMap(ECO_FACT_DERIVED_UNIT_NAMESPACE);
 102
 103                 ResultSet rs = partitioner.getResultSet();
 104
 105                 try {
 106
 107                         int i = 0;
 108
 109                         //for each common name
 110             while (rs.next()){
 111
 112                         if ((i++ % modCount) == 0 && i!= 1 ){ logger.info(pluralString + " handled: " + (i-1));}
 113
 114                         Integer taxonId = rs.getInt("IdCrrentSpec");
 115
 116
 117                                 try {
 118
 119                                         //source ref
 120                                         Reference<?> sourceRef = state.getTransactionalSourceReference();
 121
 122                                         //species
 123                                         Taxon species = createObject(rs, state);
 124
 125
 126                                         handleCountries(state, rs, species);
 127
 128                                         this.doIdCreatedUpdatedNotes(state, species, rs, taxonId, TAXON_NAMESPACE);
 129
 130                                         objectsToSave.add(species);
 131
 132
 133                                 } catch (Exception e) {
 134                                         logger.warn("Exception in current_species: IDcurrentspec " + taxonId + ". " + e.getMessage());
 135 //                                      e.printStackTrace();
 136                                 }
 137
 138             }
 139
 140 //            logger.warn("Specimen: " + countSpecimen + ", Descriptions: " + countDescriptions );
 141
 142                         logger.warn(pluralString + " to save: " + objectsToSave.size());
 143                         getTaxonService().save(objectsToSave);
 144
 145                         return success;
 146                 } catch (SQLException e) {
 147                         logger.error("SQLException:" +  e);
 148                         return false;
 149                 }
 150         }
 151
 152         private void handleCountries(GlobisImportState state, ResultSet rs, Taxon species) throws SQLException {
 153                 String countriesStr = rs.getString("dtSpcCountries");
 154                 if (isBlank(countriesStr)){
 155                         return;
 156                 }
 157                 String[] countriesSplit = countriesStr.split(";");
 158                 for (String countryStr : countriesSplit){
 159                         if (isBlank(countryStr)){
 160                                 continue;
 161                         }
 162                         countryStr = countryStr.trim();
 163
 164                         //TODO use isComplete
 165                         boolean isComplete = countryStr.endsWith(".");
 166                         if (isComplete){
 167                                 countryStr = countryStr.substring(0,countryStr.length() - 1).trim();
 168                         }
 169                         boolean isDoubtful = countryStr.endsWith("[?]");
 170                         if (isDoubtful){
 171                                 countryStr = countryStr.substring(0,countryStr.length() - 3).trim();
 172                         }
 173                         if (countryStr.startsWith("?")){
 174                                 isDoubtful = true;
 175                                 countryStr = countryStr.substring(1).trim();
 176                         }
 177
 178
 179
 180                         countryStr = normalizeCountry(countryStr);
 181
 182                         WaterbodyOrCountry country = getCountry(state, countryStr);
 183
 184                         PresenceTerm status;
 185                         if (isDoubtful){
 186                                 status = PresenceTerm.PRESENT_DOUBTFULLY();
 187                         }else{
 188                                 status = PresenceTerm.PRESENT();
 189                         }
 190
 191                         if (country != null){
 192                                 TaxonDescription desc = getTaxonDescription(species, state.getTransactionalSourceReference(), false, true);
 193                                 Distribution distribution = Distribution.NewInstance(country, status);
 194                                 desc.addElement(distribution);
 195                         }else{
 196                                 logger.warn("Country string not recognized: " + countryStr);
 197                         }
 198                 }
 199         }
 200
 201
 202
 203         /**
 204          * @param countryStr
 205          * @return
 206          */
 207         private String normalizeCountry(String countryStr) {
 208                 String result = countryStr.trim();
 209                 if (result.endsWith(".")){
 210                         result = result.substring(0,result.length() - 1);
 211                 }
 212                 return result;
 213         }
 214
 215
 216         /* (non-Javadoc)
 217          * @see eu.etaxonomy.cdm.io.common.mapping.IMappingImport#createObject(java.sql.ResultSet, eu.etaxonomy.cdm.io.common.ImportStateBase)
 218          */
 219         public Taxon createObject(ResultSet rs, GlobisImportState state)
 220                         throws SQLException {
 221                 String speciesEpi = rs.getString("dtSpcSpcakt");
 222                 String subGenusEpi = rs.getString("dtSpcSubgenakt");
 223                 String genusEpi = rs.getString("dtSpcGenusakt");
 224                 String author = rs.getString("dtSpcAutor");
 225
 226
 227                 ZoologicalName zooName = ZoologicalName.NewInstance(Rank.SPECIES());
 228                 zooName.setSpecificEpithet(speciesEpi);
 229                 if (StringUtils.isNotBlank(subGenusEpi)){
 230                         zooName.setInfraGenericEpithet(subGenusEpi);
 231                 }
 232                 zooName.setGenusOrUninomial(genusEpi);
 233                 handleAuthorAndYear(author, zooName);
 234
 235                 Taxon taxon = Taxon.NewInstance(zooName, state.getTransactionalSourceReference());
 236
 237                 return taxon;
 238         }
 239
 240
 241
 242
 243
 244         /* (non-Javadoc)
 245          * @see eu.etaxonomy.cdm.io.berlinModel.in.IPartitionedIO#getRelatedObjectsForPartition(java.sql.ResultSet)
 246          */
 247         public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs) {
 248                 String nameSpace;
 249                 Class cdmClass;
 250                 Set<String> idSet;
 251                 Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>();
 252                 try{
 253                         Set<String> taxonIdSet = new HashSet<String>();
 254
 255                         while (rs.next()){
 256 //                              handleForeignKey(rs, taxonIdSet, "taxonId");
 257                         }
 258
 259                         //taxon map
 260                         nameSpace = TAXON_NAMESPACE;
 261                         cdmClass = Taxon.class;
 262                         idSet = taxonIdSet;
 263                         Map<String, Taxon> objectMap = (Map<String, Taxon>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 264                         result.put(nameSpace, objectMap);
 265
 266
 267                 } catch (SQLException e) {
 268                         throw new RuntimeException(e);
 269                 }
 270                 return result;
 271         }
 272
 273         /* (non-Javadoc)
 274          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 275          */
 276         @Override
 277         protected boolean doCheck(GlobisImportState state){
 278                 IOValidator<GlobisImportState> validator = new GlobisCurrentSpeciesImportValidator();
 279                 return validator.validate(state);
 280         }
 281
 282
 283         /* (non-Javadoc)
 284          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 285          */
 286         protected boolean isIgnore(GlobisImportState state){
 287                 return ! state.getConfig().isDoCurrentTaxa();
 288         }
 289
 290
 291
 292
 293
 294 }