app-import/src/main/java/eu/etaxonomy/cdm/io/globis/GlobisImageImport.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.io.IOException;
  13 import java.net.MalformedURLException;
  14 import java.net.URI;
  15 import java.sql.ResultSet;
  16 import java.sql.SQLException;
  17 import java.util.HashMap;
  18 import java.util.HashSet;
  19 import java.util.Map;
  20 import java.util.Set;
  21 import java.util.UUID;
  22 import java.util.regex.Matcher;
  23 import java.util.regex.Pattern;
  24
  25 import org.apache.http.client.ClientProtocolException;
  26 import org.apache.log4j.Logger;
  27 import org.springframework.stereotype.Component;
  28
  29 import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade;
  30 import eu.etaxonomy.cdm.common.UriUtils;
  31 import eu.etaxonomy.cdm.io.common.IOValidator;
  32 import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
  33 import eu.etaxonomy.cdm.io.globis.validation.GlobisImageImportValidator;
  34 import eu.etaxonomy.cdm.model.common.Annotation;
  35 import eu.etaxonomy.cdm.model.common.CdmBase;
  36 import eu.etaxonomy.cdm.model.common.Language;
  37 import eu.etaxonomy.cdm.model.common.Marker;
  38 import eu.etaxonomy.cdm.model.common.MarkerType;
  39 import eu.etaxonomy.cdm.model.common.OriginalSourceType;
  40 import eu.etaxonomy.cdm.model.media.Media;
  41 import eu.etaxonomy.cdm.model.name.ZoologicalName;
  42 import eu.etaxonomy.cdm.model.occurrence.Collection;
  43 import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
  44 import eu.etaxonomy.cdm.model.occurrence.DeterminationEvent;
  45 import eu.etaxonomy.cdm.model.reference.Reference;
  46 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  47 import eu.etaxonomy.cdm.model.taxon.Taxon;
  48
  49
  50 /**
  51  * @author a.mueller
  52  * @created 20.02.2010
  53  * @version 1.0
  54  */
  55 @Component
  56 public class GlobisImageImport  extends GlobisImportBase<Taxon> {
  57         private static final Logger logger = Logger.getLogger(GlobisImageImport.class);
  58
  59         private int modCount = 1000;
  60
  61         private UUID uuidArtNonSpecTaxMarkerType = UUID.fromString("be362085-0f5b-4314-96d1-78b9b129ef6d") ;
  62         private static final String pluralString = "images";
  63         private static final String dbTableName = "Einzelbilder";
  64         private static final Class<?> cdmTargetClass = Media.class;  //not needed
  65
  66         private static UUID uuidGartRef = UUID.fromString("af85470f-6e54-4304-9d29-fd117cd56161");
  67
  68         public GlobisImageImport(){
  69                 super(pluralString, dbTableName, cdmTargetClass);
  70         }
  71
  72
  73
  74
  75         /* (non-Javadoc)
  76          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#getIdQuery()
  77          */
  78         @Override
  79         protected String getIdQuery() {
  80                 String strRecordQuery =
  81                         " SELECT BildId " +
  82                         " FROM " + dbTableName;
  83                 return strRecordQuery;
  84         }
  85
  86
  87
  88
  89         /* (non-Javadoc)
  90          * @see eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportBase#getRecordQuery(eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator)
  91          */
  92         @Override
  93         protected String getRecordQuery(GlobisImportConfigurator config) {
  94                 String strRecordQuery =
  95                         " SELECT i.*, NULL as Created_When, NULL as Created_Who," +
  96                                 "  NULL as Updated_who, NULL as Updated_When, NULL as Notes, st.SpecCurrspecID " +
  97                         " FROM " + getTableName() + " i " +
  98                                 " LEFT JOIN specTax st ON i.spectaxID = st.SpecTaxID " +
  99                         " WHERE ( i.BildId IN (" + ID_LIST_TOKEN + ") )";
 100                 return strRecordQuery;
 101         }
 102
 103
 104
 105         /* (non-Javadoc)
 106          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doPartition(eu.etaxonomy.cdm.io.common.ResultSetPartitioner, eu.etaxonomy.cdm.io.globis.GlobisImportState)
 107          */
 108         @Override
 109         public boolean doPartition(ResultSetPartitioner partitioner, GlobisImportState state) {
 110                 boolean success = true;
 111
 112                 Set<Media> objectsToSave = new HashSet<Media>();
 113
 114                 Map<String, DerivedUnit> typeMap = (Map<String, DerivedUnit>) partitioner.getObjectMap(TYPE_NAMESPACE);
 115
 116                 Map<String, Taxon> taxonMap = (Map<String, Taxon>) partitioner.getObjectMap(TAXON_NAMESPACE);
 117                 Map<String, ZoologicalName> specTaxNameMap = (Map<String, ZoologicalName>) partitioner.getObjectMap(SPEC_TAX_NAMESPACE);
 118
 119                 ResultSet rs = partitioner.getResultSet();
 120
 121                 Reference<?> refGart = getReferenceService().find(uuidGartRef);
 122
 123
 124                 try {
 125
 126                         int i = 0;
 127
 128                         //for each record
 129             while (rs.next()){
 130
 131                         if ((i++ % modCount) == 0 && i!= 1 ){ logger.info(pluralString + " handled: " + (i-1));}
 132
 133                         Integer bildID = rs.getInt("BildID");
 134                         Integer spectaxID = nullSafeInt(rs, "spectaxID");
 135                         Integer taxonID = nullSafeInt(rs, "SpecCurrspecID");
 136                         String copyright = rs.getString("copyright");
 137                         String specimenId = rs.getString("specimenID");
 138                         String bemerkungen = rs.getString("Bemerkungen");
 139                         String artNotSpecTax = rs.getString("Art non spectax");
 140                         String motiv = rs.getString("Motiv");
 141
 142                         //ignore:
 143                         //      [file lab2], same as Dateiname04 but less data
 144                         //      Dateipfad
 145
 146                         Set<Media> recordMedia = new HashSet<Media>();
 147
 148                         try {
 149
 150                                 makeAllMedia(state, rs, recordMedia, objectsToSave);
 151
 152                                 String title = null;
 153
 154                                 DerivedUnit specimen = null;
 155                                 if (spectaxID != null){
 156                                         //try to find type specimen
 157                                         if (isNotBlank(motiv) && (motiv.startsWith("type specimen"))){
 158                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 159                                                 String id = GlobisSpecTaxImport.getTypeId(spectaxID, collectionCode);
 160                                                 specimen = typeMap.get(id);
 161                                         }
 162
 163                                         //try to find specTaxName
 164                                         ZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 165                                 if (specTaxTaxonName != null){
 166                                         title = " taxon name " + specTaxTaxonName.getTitleCache();
 167                                 }else{
 168                                         title = " spectaxID " + spectaxID;
 169                                 }
 170                                 }else{
 171                                         title = " name " + getNameFromFileOs(rs) + (isBlank(specimenId)? "" : " (specimenId: " + specimenId + ")");
 172                                 }
 173
 174                                 //not type specimen
 175                                 if (specimen == null){
 176                                                 specimen = DerivedUnit.NewPreservedSpecimenInstance();
 177                                                 specimen.setTitleCache("Specimen for " + title );
 178                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 179                                                 //TODO
 180                                                 Collection collection = getCollection(collectionCode);
 181                                                 specimen.setCollection(collection);
 182                                         }
 183
 184
 185                                         //source
 186                                         specimen.addSource(OriginalSourceType.Import, String.valueOf(bildID), IMAGE_NAMESPACE, state.getTransactionalSourceReference(), null);
 187
 188                                         //GART id (specimenID)
 189                                         if (isNotBlank(specimenId)){
 190                                                 specimen.addSource(OriginalSourceType.Lineage, specimenId, "specimenId", refGart, null);
 191                                         }
 192                                         //bemerkungen
 193                                         if (isNotBlank(bemerkungen)){
 194                                                 Annotation annotation = Annotation.NewInstance(bemerkungen, null, null);
 195                                                 specimen.addAnnotation(annotation);
 196                                         }
 197                                         //media
 198                                         DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(specimen);
 199                                         for (Media media: recordMedia){
 200                                                 facade.addDerivedUnitMedia(media);
 201                                         }
 202                                         //art non spectax
 203                                         if (isNotBlank(artNotSpecTax)){
 204                                                 if (artNotSpecTax.equalsIgnoreCase("ja")){
 205                                                         MarkerType artNotSpecTaxMarker = getMarkerType(state, uuidArtNonSpecTaxMarkerType  , "Art non spectax", "This marker is true if in the orginal data the 'Art non spectax' was 'ja'", null) ;
 206                                                         specimen.addMarker(Marker.NewInstance(artNotSpecTaxMarker, true));
 207                                                 }else{
 208                                                         logger.warn(artNotSpecTax + " is not a valid value for 'Art non spectax' (BildID: " + bildID + ")" );
 209                                                 }
 210                                         }
 211
 212                                         if (spectaxID != null){
 213
 214                                                 //add to image gallery (discuss if this is also needed if taxon is already added to type specimen
 215 //                                              Taxon taxon = taxonMap.get(String.valueOf(taxonID));
 216                                                 ZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 217
 218 //
 219 //                                              if (taxon == null){
 220 ////                                                    taxon = specTaxMap.get(String.valueOf(spectaxID));
 221 ////                                                    specTaxName = specTaxMap.g
 222 //                                              }
 223 //                                              if (taxon == null){
 224 //                                                      logger.warn("No taxon available for specTaxID: " +  spectaxID);
 225 //                                              }else{
 226 //                                                      name = CdmBase.deproxy(taxon.getName(), ZoologicalName.class);
 227 //                                              }
 228
 229                                                 //TODO FIXME
 230
 231                                                 if (specTaxTaxonName == null){
 232                                                         logger.warn("Name could not be found for spectaxID: " + spectaxID +  " in BildID: " + bildID);
 233                                                 }else{
 234                                                         Taxon taxon = null;
 235                                                         for (Taxon specTaxTaxon: specTaxTaxonName.getTaxa()){
 236                                                                 taxon = specTaxTaxon;
 237                                                         }
 238                                                         if (taxon == null){
 239                                                                 //FIXME
 240                                                                 Reference<?> undefinedSec = null;
 241                                                                 taxon = Taxon.NewInstance(specTaxTaxonName, undefinedSec);
 242                                                         }
 243
 244                                                         DeterminationEvent.NewInstance(taxon, specimen);
 245
 246                                                 }
 247
 248
 249
 250
 251 //                                              if (taxon != null){
 252 //                                                      TaxonDescription taxonDescription = getTaxonDescription(taxon, true, true);
 253 //                                                      if (taxonDescription.getElements().size() == 0){
 254 //                                                              TextData textData = TextData.NewInstance(Feature.IMAGE());
 255 //                                                              taxonDescription.addElement(textData);
 256 //                                                      }
 257 //                                                      Set<DescriptionElementBase> elements = taxonDescription.getElements();
 258 //                                                      TextData textData = CdmBase.deproxy(elements.iterator().next(), TextData.class);
 259 //                                                      for (Media media: recordMedia){
 260 //                                                              textData.addMedia(media);
 261 //                                                      }
 262 //                                              }
 263                                         }
 264
 265                                 } catch (Exception e) {
 266                                         logger.warn("Exception in Einzelbilder: bildID " + bildID + ". " + e.getMessage());
 267                                         e.printStackTrace();
 268                                 }
 269
 270             }
 271
 272                         logger.info(pluralString + " to save: " + objectsToSave.size());
 273                         getMediaService().save(objectsToSave);
 274
 275                         return success;
 276                 } catch (SQLException e) {
 277                         logger.error("SQLException:" +  e);
 278                         return false;
 279                 }
 280         }
 281
 282         private Collection getCollection(String collectionCode) {
 283                 //TODO
 284                 return null;
 285         }
 286
 287
 288
 289
 290         private String getNameFromFileOs(ResultSet rs) throws SQLException {
 291                 String fileOS = rs.getString("file OS");
 292                 Pattern pattern = Pattern.compile("(.+)(_.{4}(-.{1,3})?(_Nr\\d{3,4})?_.{2,3}\\.jpg)");
 293                 Matcher matcher = pattern.matcher(fileOS);
 294                 if (matcher.matches()){
 295                         String match = matcher.group(1);
 296                         return match;
 297                 }else{
 298                         logger.warn("FileOS does not match: " +  fileOS);
 299                         return fileOS;
 300                 }
 301         }
 302
 303
 304
 305
 306         private void makeAllMedia(GlobisImportState state, ResultSet rs, Set<Media> recordMedia, Set<Media> objectsToSave) throws SQLException{
 307                         //make image path
 308                 String pathShort = rs.getString("Dateipfad_kurz");
 309                 String fileOS = rs.getString("file OS");
 310                 pathShort= pathShort.replace(fileOS, "");
 311                 String newPath = state.getConfig().getImageBaseUrl();
 312                 String path = pathShort.replace("image:Webversionen/", newPath);
 313
 314                 Media singleMedia = makeMedia(state, rs, "file OS", "Legende 1", path, objectsToSave );
 315                 recordMedia.add(singleMedia);
 316                 singleMedia = makeMedia(state, rs, "Dateinamen02", "Legende 2", path, objectsToSave );
 317                 recordMedia.add(singleMedia);
 318                 singleMedia = makeMedia(state, rs, "Dateinamen03", "Legende 3", path, objectsToSave );
 319                 recordMedia.add(singleMedia);
 320                 singleMedia = makeMedia(state, rs, "Dateinamen04", "Legende 4", path, objectsToSave );
 321                 recordMedia.add(singleMedia);
 322
 323         }
 324
 325         private Media makeMedia(GlobisImportState state, ResultSet rs, String fileNameAttr, String legendAttr, String path, Set<Media> objectsToSave) throws SQLException {
 326                 Media media = null;
 327                 String fileName = rs.getString(fileNameAttr);
 328                 String legend = rs.getString(legendAttr);
 329                 Integer bildID = rs.getInt("BildID");
 330
 331                 String uriStr = path+fileName;
 332                 uriStr = uriStr.replace(" ", "%20");
 333
 334                 URI uri = URI.create(uriStr);
 335
 336 //              Media media = ImageInfo.NewInstanceWithMetaData(uri, null);
 337
 338                 try {
 339                         boolean readMediaData = state.getConfig().isDoReadMediaData();
 340                         if (isBlank(legend) && readMediaData){
 341                                 if (UriUtils.isOk(UriUtils.getResponse(uri, null))){
 342                                         logger.warn("Image exists but legend is null " + uri + ", bildID" + bildID );
 343                                 }else{
 344                                         return null;
 345                                 }
 346                         }
 347
 348                         media = this.getImageMedia(uri.toString(), readMediaData);
 349                         media.putTitle(Language.ENGLISH(), legend);
 350                         this.doIdCreatedUpdatedNotes(state, media, rs, bildID, IMAGE_NAMESPACE);
 351
 352                         objectsToSave.add(media);
 353
 354
 355                 } catch (MalformedURLException e) {
 356                         e.printStackTrace();
 357                 } catch (ClientProtocolException e) {
 358                         e.printStackTrace();
 359                 } catch (IOException e) {
 360                         e.printStackTrace();
 361                 }
 362
 363                 return media;
 364         }
 365
 366         private String transformCopyright2CollectionCode(String copyright){
 367
 368                 if (isBlank(copyright)){
 369                         return "";
 370                 }else if(copyright.matches("Museum f.?r Naturkunde der Humboldt-Universit.?t, Berlin")){
 371                         return "MFNB";
 372                 }else if(copyright.matches("Staatliches Museum f.?r Tierkunde Dresden")){
 373                         return "SMTD";
 374                 }else if(copyright.equals("Natural History Museum, London")){
 375                         return "BMNH";
 376                 }else if(copyright.matches("Zoologische Staatssammlung M.?nchen")){
 377                         return "ZSSM";
 378                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Karlsruhe")){
 379                         return "SMNK";
 380                 }else if(copyright.matches("Deutsches Entomologisches Institut M.?ncheberg")){
 381                         return "DEIE";
 382                 }else if(copyright.equals("Forschungsinstitut und Naturmuseum Senckenberg")){
 383                         return "SMFM";
 384                 }else if(copyright.matches("Mus.?um National d.?Histoire Naturelle, Paris")){
 385                         return "MNHN";
 386                 }else if(copyright.equals("Naturhistorisches Museum Wien")){
 387                         return "NHMW";
 388                 }else if(copyright.equals("Naturhistoriska Riksmuseet Stockholm")){
 389                         return "NRMS";
 390                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Stuttgart")){
 391                         return "SMNS";
 392                 }else if(copyright.equals("United States National Museum of Natural History, Washington")){
 393                         return "USNM";
 394                 }else if(copyright.matches("Zentrum f.?r Biodokumentation des Saarlandes")){
 395                         return "ZFBS";
 396                 }else if(copyright.equals("Zoological Museum, University of Copenhagen")){
 397                         return "ZMUC";
 398                 }else if(copyright.equals("Zoologisches Forschungsinstitut und Museum \"Alexander Koenig\", Bonn")){
 399                         return "ZFMK";
 400                 }else if(copyright.equals("Zoologisches Forschungsmuseum \"Alexander Koenig\", Bonn")){
 401                         return "ZFMK";
 402                 }else if(copyright.matches("Zoologisches Institut der Martin-Luther-Universit.?t Halle-Wittenberg")){
 403                         return "ZIUH";
 404                 }else if(copyright.matches("Zoologisches Institut Universit.?t T.?bingen")){
 405                         return "ZIUT";
 406                 }else{
 407                         logger.warn("Unknown copyright entry: " + copyright);
 408                         return "";
 409                 }
 410         }
 411
 412
 413         @Override
 414         public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, GlobisImportState state) {
 415                 String nameSpace;
 416                 Class<?> cdmClass;
 417                 Set<String> idSet;
 418                 Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>();
 419                 try{
 420                         Set<String> currSpecIdSet = new HashSet<String>();
 421                         Set<String> specTaxIdSet = new HashSet<String>();
 422                         Set<String> typeIdSet = new HashSet<String>();
 423
 424                         while (rs.next()){
 425                                 handleForeignKey(rs, currSpecIdSet, "SpecCurrspecID");
 426                                 handleForeignKey(rs, specTaxIdSet, "spectaxID");
 427                                 handleTypeKey(rs, typeIdSet, "spectaxID", "copyright");
 428                         }
 429
 430                         //specTax map
 431                         nameSpace = SPEC_TAX_NAMESPACE;
 432                         cdmClass = ZoologicalName.class;
 433                         idSet = specTaxIdSet;
 434                         Map<String, ZoologicalName> specTaxNameMap = (Map<String, ZoologicalName>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 435                         result.put(nameSpace, specTaxNameMap);
 436
 437 //                      //taxon map
 438 //                      nameSpace = TAXON_NAMESPACE;
 439 //                      cdmClass = Taxon.class;
 440 //                      idSet = currSpecIdSet;
 441 //                      Map<String, Taxon> taxonMap = (Map<String, Taxon>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 442 //                      result.put(nameSpace, taxonMap);
 443
 444
 445                         //type map
 446                         nameSpace = GlobisSpecTaxImport.TYPE_NAMESPACE;
 447                         cdmClass = DerivedUnit.class;
 448                         idSet = typeIdSet;
 449                         Map<String, DerivedUnit> typeMap = (Map<String, DerivedUnit>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 450                         result.put(nameSpace, typeMap);
 451
 452
 453                 } catch (SQLException e) {
 454                         throw new RuntimeException(e);
 455                 }
 456                 return result;
 457         }
 458
 459         private void handleTypeKey(ResultSet rs, Set<String> idSet, String specTaxIdAttr, String copyrightAttr) throws SQLException {
 460                 Integer specTaxId = nullSafeInt(rs, specTaxIdAttr);
 461                 if (specTaxId != null){
 462                         String copyright = rs.getString(copyrightAttr);
 463                         if (isNotBlank(copyright)){
 464                                 String id  = GlobisSpecTaxImport.getTypeId(specTaxId, transformCopyright2CollectionCode(copyright));
 465                                 idSet.add(id);
 466                         }
 467                 }
 468         }
 469
 470         /* (non-Javadoc)
 471          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 472          */
 473         @Override
 474         protected boolean doCheck(GlobisImportState state){
 475                 IOValidator<GlobisImportState> validator = new GlobisImageImportValidator();
 476                 return validator.validate(state);
 477         }
 478
 479
 480         /* (non-Javadoc)
 481          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 482          */
 483         protected boolean isIgnore(GlobisImportState state){
 484                 return ! state.getConfig().isDoImages();
 485         }
 486
 487
 488
 489
 490         /* (non-Javadoc)
 491          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doInvoke(eu.etaxonomy.cdm.io.globis.GlobisImportState)
 492          */
 493         @Override
 494         protected void doInvoke(GlobisImportState state) {
 495                 Reference refGart = ReferenceFactory.newGeneric();
 496                 refGart.setTitleCache("GART");
 497                 refGart.setUuid(uuidGartRef);
 498                 getReferenceService().saveOrUpdate(refGart);
 499                 super.doInvoke(state);
 500         }
 501
 502
 503
 504
 505
 506 }