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.IZoologicalName;
  42 import eu.etaxonomy.cdm.model.name.TaxonName;
  43 import eu.etaxonomy.cdm.model.occurrence.Collection;
  44 import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
  45 import eu.etaxonomy.cdm.model.occurrence.DeterminationEvent;
  46 import eu.etaxonomy.cdm.model.reference.Reference;
  47 import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
  48 import eu.etaxonomy.cdm.model.taxon.Taxon;
  49
  50
  51 /**
  52  * @author a.mueller
  53  * @since 20.02.2010
  54  * @version 1.0
  55  */
  56 @Component
  57 public class GlobisImageImport  extends GlobisImportBase<Taxon> {
  58         private static final Logger logger = Logger.getLogger(GlobisImageImport.class);
  59
  60         private int modCount = 1000;
  61
  62         private UUID uuidArtNonSpecTaxMarkerType = UUID.fromString("be362085-0f5b-4314-96d1-78b9b129ef6d") ;
  63         private static final String pluralString = "images";
  64         private static final String dbTableName = "Einzelbilder";
  65         private static final Class<?> cdmTargetClass = Media.class;  //not needed
  66
  67         private static UUID uuidGartRef = UUID.fromString("af85470f-6e54-4304-9d29-fd117cd56161");
  68
  69         public GlobisImageImport(){
  70                 super(pluralString, dbTableName, cdmTargetClass);
  71         }
  72
  73
  74
  75
  76         /* (non-Javadoc)
  77          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#getIdQuery()
  78          */
  79         @Override
  80         protected String getIdQuery() {
  81                 String strRecordQuery =
  82                         " SELECT BildId " +
  83                         " FROM " + dbTableName;
  84                 return strRecordQuery;
  85         }
  86
  87
  88
  89
  90         /* (non-Javadoc)
  91          * @see eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportBase#getRecordQuery(eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator)
  92          */
  93         @Override
  94         protected String getRecordQuery(GlobisImportConfigurator config) {
  95                 String strRecordQuery =
  96                         " SELECT i.*, NULL as Created_When, NULL as Created_Who," +
  97                                 "  NULL as Updated_who, NULL as Updated_When, NULL as Notes, st.SpecCurrspecID " +
  98                         " FROM " + getTableName() + " i " +
  99                                 " LEFT JOIN specTax st ON i.spectaxID = st.SpecTaxID " +
 100                         " WHERE ( i.BildId IN (" + ID_LIST_TOKEN + ") )";
 101                 return strRecordQuery;
 102         }
 103
 104
 105
 106         /* (non-Javadoc)
 107          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doPartition(eu.etaxonomy.cdm.io.common.ResultSetPartitioner, eu.etaxonomy.cdm.io.globis.GlobisImportState)
 108          */
 109         @Override
 110         public boolean doPartition(ResultSetPartitioner partitioner, GlobisImportState state) {
 111                 boolean success = true;
 112
 113                 Set<Media> objectsToSave = new HashSet<>();
 114
 115                 Map<String, DerivedUnit> typeMap = partitioner.getObjectMap(TYPE_NAMESPACE);
 116
 117                 Map<String, Taxon> taxonMap = partitioner.getObjectMap(TAXON_NAMESPACE);
 118                 Map<String, TaxonName> specTaxNameMap = partitioner.getObjectMap(SPEC_TAX_NAMESPACE);
 119
 120                 ResultSet rs = partitioner.getResultSet();
 121
 122                 Reference refGart = getReferenceService().find(uuidGartRef);
 123
 124
 125                 try {
 126
 127                         int i = 0;
 128
 129                         //for each record
 130             while (rs.next()){
 131
 132                         if ((i++ % modCount) == 0 && i!= 1 ){ logger.info(pluralString + " handled: " + (i-1));}
 133
 134                         Integer bildID = rs.getInt("BildID");
 135                         Integer spectaxID = nullSafeInt(rs, "spectaxID");
 136                         Integer taxonID = nullSafeInt(rs, "SpecCurrspecID");
 137                         String copyright = rs.getString("copyright");
 138                         String specimenId = rs.getString("specimenID");
 139                         String bemerkungen = rs.getString("Bemerkungen");
 140                         String artNotSpecTax = rs.getString("Art non spectax");
 141                         String motiv = rs.getString("Motiv");
 142
 143                         //ignore:
 144                         //      [file lab2], same as Dateiname04 but less data
 145                         //      Dateipfad
 146
 147                         Set<Media> recordMedia = new HashSet<Media>();
 148
 149                         try {
 150
 151                                 makeAllMedia(state, rs, recordMedia, objectsToSave);
 152
 153                                 String title = null;
 154
 155                                 DerivedUnit specimen = null;
 156                                 if (spectaxID != null){
 157                                         //try to find type specimen
 158                                         if (isNotBlank(motiv) && (motiv.startsWith("type specimen"))){
 159                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 160                                                 String id = GlobisSpecTaxImport.getTypeId(spectaxID, collectionCode);
 161                                                 specimen = typeMap.get(id);
 162                                         }
 163
 164                                         //try to find specTaxName
 165                                         IZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 166                                 if (specTaxTaxonName != null){
 167                                         title = " taxon name " + specTaxTaxonName.getTitleCache();
 168                                 }else{
 169                                         title = " spectaxID " + spectaxID;
 170                                 }
 171                                 }else{
 172                                         title = " name " + getNameFromFileOs(rs) + (isBlank(specimenId)? "" : " (specimenId: " + specimenId + ")");
 173                                 }
 174
 175                                 //not type specimen
 176                                 if (specimen == null){
 177                                                 specimen = DerivedUnit.NewPreservedSpecimenInstance();
 178                                                 specimen.setTitleCache("Specimen for " + title );
 179                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 180                                                 //TODO
 181                                                 Collection collection = getCollection(collectionCode);
 182                                                 specimen.setCollection(collection);
 183                                         }
 184
 185
 186                                         //source
 187                                         specimen.addSource(OriginalSourceType.Import, String.valueOf(bildID), IMAGE_NAMESPACE, state.getTransactionalSourceReference(), null);
 188
 189                                         //GART id (specimenID)
 190                                         if (isNotBlank(specimenId)){
 191                                                 specimen.addSource(OriginalSourceType.Lineage, specimenId, "specimenId", refGart, null);
 192                                         }
 193                                         //bemerkungen
 194                                         if (isNotBlank(bemerkungen)){
 195                                                 Annotation annotation = Annotation.NewInstance(bemerkungen, null, null);
 196                                                 specimen.addAnnotation(annotation);
 197                                         }
 198                                         //media
 199                                         DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(specimen);
 200                                         for (Media media: recordMedia){
 201                                                 facade.addDerivedUnitMedia(media);
 202                                         }
 203                                         //art non spectax
 204                                         if (isNotBlank(artNotSpecTax)){
 205                                                 if (artNotSpecTax.equalsIgnoreCase("ja")){
 206                                                         MarkerType artNotSpecTaxMarker = getMarkerType(state, uuidArtNonSpecTaxMarkerType  , "Art non spectax", "This marker is true if in the orginal data the 'Art non spectax' was 'ja'", null) ;
 207                                                         specimen.addMarker(Marker.NewInstance(artNotSpecTaxMarker, true));
 208                                                 }else{
 209                                                         logger.warn(artNotSpecTax + " is not a valid value for 'Art non spectax' (BildID: " + bildID + ")" );
 210                                                 }
 211                                         }
 212
 213                                         if (spectaxID != null){
 214
 215                                                 //add to image gallery (discuss if this is also needed if taxon is already added to type specimen
 216 //                                              Taxon taxon = taxonMap.get(String.valueOf(taxonID));
 217                                                 IZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 218
 219 //
 220 //                                              if (taxon == null){
 221 ////                                                    taxon = specTaxMap.get(String.valueOf(spectaxID));
 222 ////                                                    specTaxName = specTaxMap.g
 223 //                                              }
 224 //                                              if (taxon == null){
 225 //                                                      logger.warn("No taxon available for specTaxID: " +  spectaxID);
 226 //                                              }else{
 227 //                                                      name = CdmBase.deproxy(taxon.getName(), ZoologicalName.class);
 228 //                                              }
 229
 230                                                 //TODO FIXME
 231
 232                                                 if (specTaxTaxonName == null){
 233                                                         logger.warn("Name could not be found for spectaxID: " + spectaxID +  " in BildID: " + bildID);
 234                                                 }else{
 235                                                         Taxon taxon = null;
 236                                                         for (Taxon specTaxTaxon: specTaxTaxonName.getTaxa()){
 237                                                                 taxon = specTaxTaxon;
 238                                                         }
 239                                                         if (taxon == null){
 240                                                                 //FIXME
 241                                                                 Reference undefinedSec = null;
 242                                                                 taxon = Taxon.NewInstance(specTaxTaxonName, undefinedSec);
 243                                                         }
 244
 245                                                         DeterminationEvent.NewInstance(taxon, specimen);
 246
 247                                                 }
 248
 249
 250
 251
 252 //                                              if (taxon != null){
 253 //                                                      TaxonDescription taxonDescription = getTaxonDescription(taxon, true, true);
 254 //                                                      if (taxonDescription.getElements().size() == 0){
 255 //                                                              TextData textData = TextData.NewInstance(Feature.IMAGE());
 256 //                                                              taxonDescription.addElement(textData);
 257 //                                                      }
 258 //                                                      Set<DescriptionElementBase> elements = taxonDescription.getElements();
 259 //                                                      TextData textData = CdmBase.deproxy(elements.iterator().next(), TextData.class);
 260 //                                                      for (Media media: recordMedia){
 261 //                                                              textData.addMedia(media);
 262 //                                                      }
 263 //                                              }
 264                                         }
 265
 266                                 } catch (Exception e) {
 267                                         logger.warn("Exception in Einzelbilder: bildID " + bildID + ". " + e.getMessage());
 268                                         e.printStackTrace();
 269                                 }
 270
 271             }
 272
 273                         logger.info(pluralString + " to save: " + objectsToSave.size());
 274                         getMediaService().save(objectsToSave);
 275
 276                         return success;
 277                 } catch (SQLException e) {
 278                         logger.error("SQLException:" +  e);
 279                         return false;
 280                 }
 281         }
 282
 283         private Collection getCollection(String collectionCode) {
 284                 //TODO
 285                 return null;
 286         }
 287
 288
 289
 290
 291         private String getNameFromFileOs(ResultSet rs) throws SQLException {
 292                 String fileOS = rs.getString("file OS");
 293                 Pattern pattern = Pattern.compile("(.+)(_.{4}(-.{1,3})?(_Nr\\d{3,4})?_.{2,3}\\.jpg)");
 294                 Matcher matcher = pattern.matcher(fileOS);
 295                 if (matcher.matches()){
 296                         String match = matcher.group(1);
 297                         return match;
 298                 }else{
 299                         logger.warn("FileOS does not match: " +  fileOS);
 300                         return fileOS;
 301                 }
 302         }
 303
 304
 305
 306
 307         private void makeAllMedia(GlobisImportState state, ResultSet rs, Set<Media> recordMedia, Set<Media> objectsToSave) throws SQLException{
 308                         //make image path
 309                 String pathShort = rs.getString("Dateipfad_kurz");
 310                 String fileOS = rs.getString("file OS");
 311                 pathShort= pathShort.replace(fileOS, "");
 312                 String newPath = state.getConfig().getImageBaseUrl();
 313                 String path = pathShort.replace("image:Webversionen/", newPath);
 314
 315                 Media singleMedia = makeMedia(state, rs, "file OS", "Legende 1", path, objectsToSave );
 316                 recordMedia.add(singleMedia);
 317                 singleMedia = makeMedia(state, rs, "Dateinamen02", "Legende 2", path, objectsToSave );
 318                 recordMedia.add(singleMedia);
 319                 singleMedia = makeMedia(state, rs, "Dateinamen03", "Legende 3", path, objectsToSave );
 320                 recordMedia.add(singleMedia);
 321                 singleMedia = makeMedia(state, rs, "Dateinamen04", "Legende 4", path, objectsToSave );
 322                 recordMedia.add(singleMedia);
 323
 324         }
 325
 326         private Media makeMedia(GlobisImportState state, ResultSet rs, String fileNameAttr, String legendAttr, String path, Set<Media> objectsToSave) throws SQLException {
 327                 Media media = null;
 328                 String fileName = rs.getString(fileNameAttr);
 329                 String legend = rs.getString(legendAttr);
 330                 Integer bildID = rs.getInt("BildID");
 331
 332                 String uriStr = path+fileName;
 333                 uriStr = uriStr.replace(" ", "%20");
 334
 335                 URI uri = URI.create(uriStr);
 336
 337 //              Media media = ImageInfo.NewInstanceWithMetaData(uri, null);
 338
 339                 try {
 340                         boolean readMediaData = state.getConfig().isDoReadMediaData();
 341                         if (isBlank(legend) && readMediaData){
 342                                 if (UriUtils.isOk(UriUtils.getResponse(uri, null))){
 343                                         logger.warn("Image exists but legend is null " + uri + ", bildID" + bildID );
 344                                 }else{
 345                                         return null;
 346                                 }
 347                         }
 348
 349                         media = this.getImageMedia(uri.toString(), readMediaData);
 350                         media.putTitle(Language.ENGLISH(), legend);
 351                         this.doIdCreatedUpdatedNotes(state, media, rs, bildID, IMAGE_NAMESPACE);
 352
 353                         objectsToSave.add(media);
 354
 355
 356                 } catch (MalformedURLException e) {
 357                         e.printStackTrace();
 358                 } catch (ClientProtocolException e) {
 359                         e.printStackTrace();
 360                 } catch (IOException e) {
 361                         e.printStackTrace();
 362                 }
 363
 364                 return media;
 365         }
 366
 367         private String transformCopyright2CollectionCode(String copyright){
 368
 369                 if (isBlank(copyright)){
 370                         return "";
 371                 }else if(copyright.matches("Museum f.?r Naturkunde der Humboldt-Universit.?t, Berlin")){
 372                         return "MFNB";
 373                 }else if(copyright.matches("Staatliches Museum f.?r Tierkunde Dresden")){
 374                         return "SMTD";
 375                 }else if(copyright.equals("Natural History Museum, London")){
 376                         return "BMNH";
 377                 }else if(copyright.matches("Zoologische Staatssammlung M.?nchen")){
 378                         return "ZSSM";
 379                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Karlsruhe")){
 380                         return "SMNK";
 381                 }else if(copyright.matches("Deutsches Entomologisches Institut M.?ncheberg")){
 382                         return "DEIE";
 383                 }else if(copyright.equals("Forschungsinstitut und Naturmuseum Senckenberg")){
 384                         return "SMFM";
 385                 }else if(copyright.matches("Mus.?um National d.?Histoire Naturelle, Paris")){
 386                         return "MNHN";
 387                 }else if(copyright.equals("Naturhistorisches Museum Wien")){
 388                         return "NHMW";
 389                 }else if(copyright.equals("Naturhistoriska Riksmuseet Stockholm")){
 390                         return "NRMS";
 391                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Stuttgart")){
 392                         return "SMNS";
 393                 }else if(copyright.equals("United States National Museum of Natural History, Washington")){
 394                         return "USNM";
 395                 }else if(copyright.matches("Zentrum f.?r Biodokumentation des Saarlandes")){
 396                         return "ZFBS";
 397                 }else if(copyright.equals("Zoological Museum, University of Copenhagen")){
 398                         return "ZMUC";
 399                 }else if(copyright.equals("Zoologisches Forschungsinstitut und Museum \"Alexander Koenig\", Bonn")){
 400                         return "ZFMK";
 401                 }else if(copyright.equals("Zoologisches Forschungsmuseum \"Alexander Koenig\", Bonn")){
 402                         return "ZFMK";
 403                 }else if(copyright.matches("Zoologisches Institut der Martin-Luther-Universit.?t Halle-Wittenberg")){
 404                         return "ZIUH";
 405                 }else if(copyright.matches("Zoologisches Institut Universit.?t T.?bingen")){
 406                         return "ZIUT";
 407                 }else{
 408                         logger.warn("Unknown copyright entry: " + copyright);
 409                         return "";
 410                 }
 411         }
 412
 413
 414         @Override
 415         public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, GlobisImportState state) {
 416                 String nameSpace;
 417                 Class<?> cdmClass;
 418                 Set<String> idSet;
 419                 Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>();
 420                 try{
 421                         Set<String> currSpecIdSet = new HashSet<>();
 422                         Set<String> specTaxIdSet = new HashSet<>();
 423                         Set<String> typeIdSet = new HashSet<>();
 424
 425                         while (rs.next()){
 426                                 handleForeignKey(rs, currSpecIdSet, "SpecCurrspecID");
 427                                 handleForeignKey(rs, specTaxIdSet, "spectaxID");
 428                                 handleTypeKey(rs, typeIdSet, "spectaxID", "copyright");
 429                         }
 430
 431                         //specTax map
 432                         nameSpace = SPEC_TAX_NAMESPACE;
 433                         cdmClass = TaxonName.class;
 434                         idSet = specTaxIdSet;
 435                         Map<String, TaxonName> specTaxNameMap = (Map<String, TaxonName>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 436                         result.put(nameSpace, specTaxNameMap);
 437
 438 //                      //taxon map
 439 //                      nameSpace = TAXON_NAMESPACE;
 440 //                      cdmClass = Taxon.class;
 441 //                      idSet = currSpecIdSet;
 442 //                      Map<String, Taxon> taxonMap = (Map<String, Taxon>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 443 //                      result.put(nameSpace, taxonMap);
 444
 445
 446                         //type map
 447                         nameSpace = GlobisSpecTaxImport.TYPE_NAMESPACE;
 448                         cdmClass = DerivedUnit.class;
 449                         idSet = typeIdSet;
 450                         Map<String, DerivedUnit> typeMap = (Map<String, DerivedUnit>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
 451                         result.put(nameSpace, typeMap);
 452
 453
 454                 } catch (SQLException e) {
 455                         throw new RuntimeException(e);
 456                 }
 457                 return result;
 458         }
 459
 460         private void handleTypeKey(ResultSet rs, Set<String> idSet, String specTaxIdAttr, String copyrightAttr) throws SQLException {
 461                 Integer specTaxId = nullSafeInt(rs, specTaxIdAttr);
 462                 if (specTaxId != null){
 463                         String copyright = rs.getString(copyrightAttr);
 464                         if (isNotBlank(copyright)){
 465                                 String id  = GlobisSpecTaxImport.getTypeId(specTaxId, transformCopyright2CollectionCode(copyright));
 466                                 idSet.add(id);
 467                         }
 468                 }
 469         }
 470
 471         /* (non-Javadoc)
 472          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 473          */
 474         @Override
 475         protected boolean doCheck(GlobisImportState state){
 476                 IOValidator<GlobisImportState> validator = new GlobisImageImportValidator();
 477                 return validator.validate(state);
 478         }
 479
 480
 481         /* (non-Javadoc)
 482          * @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
 483          */
 484         @Override
 485     protected boolean isIgnore(GlobisImportState state){
 486                 return ! state.getConfig().isDoImages();
 487         }
 488
 489
 490
 491
 492         /* (non-Javadoc)
 493          * @see eu.etaxonomy.cdm.io.globis.GlobisImportBase#doInvoke(eu.etaxonomy.cdm.io.globis.GlobisImportState)
 494          */
 495         @Override
 496         protected void doInvoke(GlobisImportState state) {
 497                 Reference refGart = ReferenceFactory.newGeneric();
 498                 refGart.setTitleCache("GART");
 499                 refGart.setUuid(uuidGartRef);
 500                 getReferenceService().saveOrUpdate(refGart);
 501                 super.doInvoke(state);
 502         }
 503
 504
 505
 506
 507
 508 }