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 eu.etaxonomy.cdm.common.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.media.Media;
  40 import eu.etaxonomy.cdm.model.name.IZoologicalName;
  41 import eu.etaxonomy.cdm.model.name.TaxonName;
  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.OriginalSourceType;
  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  */
  55 @Component
  56 public class GlobisImageImport  extends GlobisImportBase<Taxon> {
  57
  58     private static final long serialVersionUID = 5697033145326415146L;
  59     private static final Logger logger = Logger.getLogger(GlobisImageImport.class);
  60
  61         private int modCount = 1000;
  62
  63         private UUID uuidArtNonSpecTaxMarkerType = UUID.fromString("be362085-0f5b-4314-96d1-78b9b129ef6d") ;
  64         private static final String pluralString = "images";
  65         private static final String dbTableName = "Einzelbilder";
  66         private static final Class<?> cdmTargetClass = Media.class;  //not needed
  67
  68         private static UUID uuidGartRef = UUID.fromString("af85470f-6e54-4304-9d29-fd117cd56161");
  69
  70         public GlobisImageImport(){
  71                 super(pluralString, dbTableName, cdmTargetClass);
  72         }
  73
  74         @Override
  75         protected String getIdQuery() {
  76                 String strRecordQuery =
  77                         " SELECT BildId " +
  78                         " FROM " + dbTableName;
  79                 return strRecordQuery;
  80         }
  81
  82         @Override
  83         protected String getRecordQuery(GlobisImportConfigurator config) {
  84                 String strRecordQuery =
  85                         " SELECT i.*, NULL as Created_When, NULL as Created_Who," +
  86                                 "  NULL as Updated_who, NULL as Updated_When, NULL as Notes, st.SpecCurrspecID " +
  87                         " FROM " + getTableName() + " i " +
  88                                 " LEFT JOIN specTax st ON i.spectaxID = st.SpecTaxID " +
  89                         " WHERE ( i.BildId IN (" + ID_LIST_TOKEN + ") )";
  90                 return strRecordQuery;
  91         }
  92
  93         @Override
  94         public boolean doPartition(@SuppressWarnings("rawtypes") ResultSetPartitioner partitioner, GlobisImportState state) {
  95                 boolean success = true;
  96
  97                 Set<Media> objectsToSave = new HashSet<>();
  98
  99                 @SuppressWarnings("unchecked")
 100         Map<String, DerivedUnit> typeMap = partitioner.getObjectMap(TYPE_NAMESPACE);
 101
 102                 @SuppressWarnings("unchecked")
 103         Map<String, Taxon> taxonMap = partitioner.getObjectMap(TAXON_NAMESPACE);
 104                 @SuppressWarnings("unchecked")
 105         Map<String, TaxonName> specTaxNameMap = partitioner.getObjectMap(SPEC_TAX_NAMESPACE);
 106
 107                 ResultSet rs = partitioner.getResultSet();
 108
 109                 Reference refGart = getReferenceService().find(uuidGartRef);
 110
 111
 112                 try {
 113
 114                         int i = 0;
 115
 116                         //for each record
 117             while (rs.next()){
 118
 119                         if ((i++ % modCount) == 0 && i!= 1 ){ logger.info(pluralString + " handled: " + (i-1));}
 120
 121                         Integer bildID = rs.getInt("BildID");
 122                         Integer spectaxID = nullSafeInt(rs, "spectaxID");
 123                         Integer taxonID = nullSafeInt(rs, "SpecCurrspecID");
 124                         String copyright = rs.getString("copyright");
 125                         String specimenId = rs.getString("specimenID");
 126                         String bemerkungen = rs.getString("Bemerkungen");
 127                         String artNotSpecTax = rs.getString("Art non spectax");
 128                         String motiv = rs.getString("Motiv");
 129
 130                         //ignore:
 131                         //      [file lab2], same as Dateiname04 but less data
 132                         //      Dateipfad
 133
 134                         Set<Media> recordMedia = new HashSet<>();
 135
 136                         try {
 137
 138                                 makeAllMedia(state, rs, recordMedia, objectsToSave);
 139
 140                                 String title = null;
 141
 142                                 DerivedUnit specimen = null;
 143                                 if (spectaxID != null){
 144                                         //try to find type specimen
 145                                         if (isNotBlank(motiv) && (motiv.startsWith("type specimen"))){
 146                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 147                                                 String id = GlobisSpecTaxImport.getTypeId(spectaxID, collectionCode);
 148                                                 specimen = typeMap.get(id);
 149                                         }
 150
 151                                         //try to find specTaxName
 152                                         IZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 153                                 if (specTaxTaxonName != null){
 154                                         title = " taxon name " + specTaxTaxonName.getTitleCache();
 155                                 }else{
 156                                         title = " spectaxID " + spectaxID;
 157                                 }
 158                                 }else{
 159                                         title = " name " + getNameFromFileOs(rs) + (isBlank(specimenId)? "" : " (specimenId: " + specimenId + ")");
 160                                 }
 161
 162                                 //not type specimen
 163                                 if (specimen == null){
 164                                                 specimen = DerivedUnit.NewPreservedSpecimenInstance();
 165                                                 specimen.setTitleCache("Specimen for " + title, true);
 166                                                 String collectionCode = transformCopyright2CollectionCode(copyright);
 167                                                 //TODO
 168                                                 Collection collection = getCollection(collectionCode);
 169                                                 specimen.setCollection(collection);
 170                                         }
 171
 172                                         //source
 173                                         specimen.addSource(OriginalSourceType.Import, String.valueOf(bildID), IMAGE_NAMESPACE, state.getTransactionalSourceReference(), null);
 174
 175                                         //GART id (specimenID)
 176                                         if (isNotBlank(specimenId)){
 177                                                 specimen.addSource(OriginalSourceType.Lineage, specimenId, "specimenId", refGart, null);
 178                                         }
 179                                         //bemerkungen
 180                                         if (isNotBlank(bemerkungen)){
 181                                                 Annotation annotation = Annotation.NewInstance(bemerkungen, null, null);
 182                                                 specimen.addAnnotation(annotation);
 183                                         }
 184                                         //media
 185                                         DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(specimen);
 186                                         for (Media media: recordMedia){
 187                                                 facade.addDerivedUnitMedia(media);
 188                                         }
 189                                         //art non spectax
 190                                         if (isNotBlank(artNotSpecTax)){
 191                                                 if (artNotSpecTax.equalsIgnoreCase("ja")){
 192                                                         MarkerType artNotSpecTaxMarker = getMarkerType(state, uuidArtNonSpecTaxMarkerType  , "Art non spectax", "This marker is true if in the orginal data the 'Art non spectax' was 'ja'", null) ;
 193                                                         specimen.addMarker(Marker.NewInstance(artNotSpecTaxMarker, true));
 194                                                 }else{
 195                                                         logger.warn(artNotSpecTax + " is not a valid value for 'Art non spectax' (BildID: " + bildID + ")" );
 196                                                 }
 197                                         }
 198
 199                                         if (spectaxID != null){
 200
 201                                                 //add to image gallery (discuss if this is also needed if taxon is already added to type specimen
 202 //                                              Taxon taxon = taxonMap.get(String.valueOf(taxonID));
 203                                                 IZoologicalName specTaxTaxonName = specTaxNameMap.get(String.valueOf(spectaxID));
 204
 205 //
 206 //                                              if (taxon == null){
 207 ////                                                    taxon = specTaxMap.get(String.valueOf(spectaxID));
 208 ////                                                    specTaxName = specTaxMap.g
 209 //                                              }
 210 //                                              if (taxon == null){
 211 //                                                      logger.warn("No taxon available for specTaxID: " +  spectaxID);
 212 //                                              }else{
 213 //                                                      name = CdmBase.deproxy(taxon.getName(), ZoologicalName.class);
 214 //                                              }
 215
 216                                                 //TODO FIXME
 217
 218                                                 if (specTaxTaxonName == null){
 219                                                         logger.warn("Name could not be found for spectaxID: " + spectaxID +  " in BildID: " + bildID);
 220                                                 }else{
 221                                                         Taxon taxon = null;
 222                                                         for (Taxon specTaxTaxon: specTaxTaxonName.getTaxa()){
 223                                                                 taxon = specTaxTaxon;
 224                                                         }
 225                                                         if (taxon == null){
 226                                                                 //FIXME
 227                                                                 Reference undefinedSec = null;
 228                                                                 taxon = Taxon.NewInstance(specTaxTaxonName, undefinedSec);
 229                                                         }
 230
 231                                                         DeterminationEvent.NewInstance(taxon, specimen);
 232
 233                                                 }
 234
 235
 236
 237
 238 //                                              if (taxon != null){
 239 //                                                      TaxonDescription taxonDescription = getTaxonDescription(taxon, true, true);
 240 //                                                      if (taxonDescription.getElements().size() == 0){
 241 //                                                              TextData textData = TextData.NewInstance(Feature.IMAGE());
 242 //                                                              taxonDescription.addElement(textData);
 243 //                                                      }
 244 //                                                      Set<DescriptionElementBase> elements = taxonDescription.getElements();
 245 //                                                      TextData textData = CdmBase.deproxy(elements.iterator().next(), TextData.class);
 246 //                                                      for (Media media: recordMedia){
 247 //                                                              textData.addMedia(media);
 248 //                                                      }
 249 //                                              }
 250                                         }
 251
 252                                 } catch (Exception e) {
 253                                         logger.warn("Exception in Einzelbilder: bildID " + bildID + ". " + e.getMessage());
 254                                         e.printStackTrace();
 255                                 }
 256
 257             }
 258
 259                         logger.info(pluralString + " to save: " + objectsToSave.size());
 260                         getMediaService().save(objectsToSave);
 261
 262                         return success;
 263                 } catch (SQLException e) {
 264                         logger.error("SQLException:" +  e);
 265                         return false;
 266                 }
 267         }
 268
 269         private Collection getCollection(String collectionCode) {
 270                 //TODO
 271                 return null;
 272         }
 273
 274         private String getNameFromFileOs(ResultSet rs) throws SQLException {
 275                 String fileOS = rs.getString("file OS");
 276                 Pattern pattern = Pattern.compile("(.+)(_.{4}(-.{1,3})?(_Nr\\d{3,4})?_.{2,3}\\.jpg)");
 277                 Matcher matcher = pattern.matcher(fileOS);
 278                 if (matcher.matches()){
 279                         String match = matcher.group(1);
 280                         return match;
 281                 }else{
 282                         logger.warn("FileOS does not match: " +  fileOS);
 283                         return fileOS;
 284                 }
 285         }
 286
 287         private void makeAllMedia(GlobisImportState state, ResultSet rs, Set<Media> recordMedia, Set<Media> objectsToSave) throws SQLException{
 288                         //make image path
 289                 String pathShort = rs.getString("Dateipfad_kurz");
 290                 String fileOS = rs.getString("file OS");
 291                 pathShort= pathShort.replace(fileOS, "");
 292                 String newPath = state.getConfig().getImageBaseUrl();
 293                 String path = pathShort.replace("image:Webversionen/", newPath);
 294
 295                 Media singleMedia = makeMedia(state, rs, "file OS", "Legende 1", path, objectsToSave );
 296                 recordMedia.add(singleMedia);
 297                 singleMedia = makeMedia(state, rs, "Dateinamen02", "Legende 2", path, objectsToSave );
 298                 recordMedia.add(singleMedia);
 299                 singleMedia = makeMedia(state, rs, "Dateinamen03", "Legende 3", path, objectsToSave );
 300                 recordMedia.add(singleMedia);
 301                 singleMedia = makeMedia(state, rs, "Dateinamen04", "Legende 4", path, objectsToSave );
 302                 recordMedia.add(singleMedia);
 303
 304         }
 305
 306         private Media makeMedia(GlobisImportState state, ResultSet rs, String fileNameAttr, String legendAttr, String path, Set<Media> objectsToSave) throws SQLException {
 307                 Media media = null;
 308                 String fileName = rs.getString(fileNameAttr);
 309                 String legend = rs.getString(legendAttr);
 310                 Integer bildID = rs.getInt("BildID");
 311
 312                 String uriStr = path+fileName;
 313                 uriStr = uriStr.replace(" ", "%20");
 314
 315                 URI uri = URI.create(uriStr);
 316
 317 //              Media media = ImageInfo.NewInstanceWithMetaData(uri, null);
 318
 319                 try {
 320                         boolean readMediaData = state.getConfig().isDoReadMediaData();
 321                         if (isBlank(legend) && readMediaData){
 322                                 if (UriUtils.isOk(UriUtils.getResponse(uri, null))){
 323                                         logger.warn("Image exists but legend is null " + uri + ", bildID" + bildID );
 324                                 }else{
 325                                         return null;
 326                                 }
 327                         }
 328
 329                         media = this.getImageMedia(uri.toString(), readMediaData);
 330                         media.putTitle(Language.ENGLISH(), legend);
 331                         this.doIdCreatedUpdatedNotes(state, media, rs, bildID, IMAGE_NAMESPACE);
 332
 333                         objectsToSave.add(media);
 334
 335
 336                 } catch (MalformedURLException e) {
 337                         e.printStackTrace();
 338                 } catch (ClientProtocolException e) {
 339                         e.printStackTrace();
 340                 } catch (IOException e) {
 341                         e.printStackTrace();
 342                 }
 343
 344                 return media;
 345         }
 346
 347         private String transformCopyright2CollectionCode(String copyright){
 348
 349                 if (isBlank(copyright)){
 350                         return "";
 351                 }else if(copyright.matches("Museum f.?r Naturkunde der Humboldt-Universit.?t, Berlin")){
 352                         return "MFNB";
 353                 }else if(copyright.matches("Staatliches Museum f.?r Tierkunde Dresden")){
 354                         return "SMTD";
 355                 }else if(copyright.equals("Natural History Museum, London")){
 356                         return "BMNH";
 357                 }else if(copyright.matches("Zoologische Staatssammlung M.?nchen")){
 358                         return "ZSSM";
 359                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Karlsruhe")){
 360                         return "SMNK";
 361                 }else if(copyright.matches("Deutsches Entomologisches Institut M.?ncheberg")){
 362                         return "DEIE";
 363                 }else if(copyright.equals("Forschungsinstitut und Naturmuseum Senckenberg")){
 364                         return "SMFM";
 365                 }else if(copyright.matches("Mus.?um National d.?Histoire Naturelle, Paris")){
 366                         return "MNHN";
 367                 }else if(copyright.equals("Naturhistorisches Museum Wien")){
 368                         return "NHMW";
 369                 }else if(copyright.equals("Naturhistoriska Riksmuseet Stockholm")){
 370                         return "NRMS";
 371                 }else if(copyright.matches("Staatliches Museum f.?r Naturkunde Stuttgart")){
 372                         return "SMNS";
 373                 }else if(copyright.equals("United States National Museum of Natural History, Washington")){
 374                         return "USNM";
 375                 }else if(copyright.matches("Zentrum f.?r Biodokumentation des Saarlandes")){
 376                         return "ZFBS";
 377                 }else if(copyright.equals("Zoological Museum, University of Copenhagen")){
 378                         return "ZMUC";
 379                 }else if(copyright.equals("Zoologisches Forschungsinstitut und Museum \"Alexander Koenig\", Bonn")){
 380                         return "ZFMK";
 381                 }else if(copyright.equals("Zoologisches Forschungsmuseum \"Alexander Koenig\", Bonn")){
 382                         return "ZFMK";
 383                 }else if(copyright.matches("Zoologisches Institut der Martin-Luther-Universit.?t Halle-Wittenberg")){
 384                         return "ZIUH";
 385                 }else if(copyright.matches("Zoologisches Institut Universit.?t T.?bingen")){
 386                         return "ZIUT";
 387                 }else{
 388                         logger.warn("Unknown copyright entry: " + copyright);
 389                         return "";
 390                 }
 391         }
 392
 393
 394         @Override
 395         public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, GlobisImportState state) {
 396
 397             String nameSpace;
 398                 Set<String> idSet;
 399                 Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
 400                 try{
 401                         Set<String> currSpecIdSet = new HashSet<>();
 402                         Set<String> specTaxIdSet = new HashSet<>();
 403                         Set<String> typeIdSet = new HashSet<>();
 404
 405                         while (rs.next()){
 406                                 handleForeignKey(rs, currSpecIdSet, "SpecCurrspecID");
 407                                 handleForeignKey(rs, specTaxIdSet, "spectaxID");
 408                                 handleTypeKey(rs, typeIdSet, "spectaxID", "copyright");
 409                         }
 410
 411                         //specTax map
 412                         nameSpace = SPEC_TAX_NAMESPACE;
 413                         idSet = specTaxIdSet;
 414                         Map<String, TaxonName> specTaxNameMap = getCommonService().getSourcedObjectsByIdInSourceC(TaxonName.class, idSet, nameSpace);
 415                         result.put(nameSpace, specTaxNameMap);
 416
 417 //                      //taxon map
 418 //                      nameSpace = TAXON_NAMESPACE;
 419 //                      idSet = currSpecIdSet;
 420 //                      Map<String, Taxon> taxonMap = getCommonService().getSourcedObjectsByIdInSourceC(Taxon.class, idSet, nameSpace);
 421 //                      result.put(nameSpace, taxonMap);
 422
 423
 424                         //type map
 425                         nameSpace = GlobisImportBase.TYPE_NAMESPACE;
 426                         idSet = typeIdSet;
 427                         Map<String, DerivedUnit> typeMap = getCommonService().getSourcedObjectsByIdInSourceC(DerivedUnit.class, idSet, nameSpace);
 428                         result.put(nameSpace, typeMap);
 429
 430
 431                 } catch (SQLException e) {
 432                         throw new RuntimeException(e);
 433                 }
 434                 return result;
 435         }
 436
 437         private void handleTypeKey(ResultSet rs, Set<String> idSet, String specTaxIdAttr, String copyrightAttr) throws SQLException {
 438                 Integer specTaxId = nullSafeInt(rs, specTaxIdAttr);
 439                 if (specTaxId != null){
 440                         String copyright = rs.getString(copyrightAttr);
 441                         if (isNotBlank(copyright)){
 442                                 String id  = GlobisSpecTaxImport.getTypeId(specTaxId, transformCopyright2CollectionCode(copyright));
 443                                 idSet.add(id);
 444                         }
 445                 }
 446         }
 447
 448         @Override
 449         protected boolean doCheck(GlobisImportState state){
 450                 IOValidator<GlobisImportState> validator = new GlobisImageImportValidator();
 451                 return validator.validate(state);
 452         }
 453
 454         @Override
 455     protected boolean isIgnore(GlobisImportState state){
 456                 return ! state.getConfig().isDoImages();
 457         }
 458
 459         @Override
 460         protected void doInvoke(GlobisImportState state) {
 461                 Reference refGart = ReferenceFactory.newGeneric();
 462                 refGart.setTitleCache("GART", true);
 463                 refGart.setUuid(uuidGartRef);
 464                 getReferenceService().saveOrUpdate(refGart);
 465                 super.doInvoke(state);
 466         }
 467 }