Revision f2be28b3
Added by Andreas Müller over 11 years ago
| app-import/src/main/java/eu/etaxonomy/cdm/io/globis/GlobisSpecTaxImport.java | ||
|---|---|---|
| 13 | 13 |
import java.sql.SQLException; |
| 14 | 14 |
import java.util.HashMap; |
| 15 | 15 |
import java.util.HashSet; |
| 16 |
import java.util.List; |
|
| 16 | 17 |
import java.util.Map; |
| 17 | 18 |
import java.util.Set; |
| 18 | 19 |
import java.util.regex.Matcher; |
| 19 | 20 |
import java.util.regex.Pattern; |
| 20 | 21 |
|
| 21 |
import org.apache.commons.lang.StringUtils; |
|
| 22 | 22 |
import org.apache.log4j.Logger; |
| 23 | 23 |
import org.springframework.stereotype.Component; |
| 24 | 24 |
|
| 25 | 25 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade; |
| 26 | 26 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade.DerivedUnitType; |
| 27 | 27 |
import eu.etaxonomy.cdm.common.CdmUtils; |
| 28 |
import eu.etaxonomy.cdm.io.common.IImportConfigurator; |
|
| 29 | 28 |
import eu.etaxonomy.cdm.io.common.IOValidator; |
| 30 | 29 |
import eu.etaxonomy.cdm.io.common.ResultSetPartitioner; |
| 31 | 30 |
import eu.etaxonomy.cdm.io.common.mapping.IMappingImport; |
| 32 |
import eu.etaxonomy.cdm.io.globis.validation.GlobisReferenceImportValidator; |
|
| 33 | 31 |
import eu.etaxonomy.cdm.io.globis.validation.GlobisSpecTaxaImportValidator; |
| 34 | 32 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
| 35 |
import eu.etaxonomy.cdm.model.common.Extension; |
|
| 36 | 33 |
import eu.etaxonomy.cdm.model.common.IdentifiableSource; |
| 37 | 34 |
import eu.etaxonomy.cdm.model.common.Marker; |
| 38 | 35 |
import eu.etaxonomy.cdm.model.common.MarkerType; |
| ... | ... | |
| 41 | 38 |
import eu.etaxonomy.cdm.model.name.Rank; |
| 42 | 39 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignation; |
| 43 | 40 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus; |
| 44 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationTest; |
|
| 45 | 41 |
import eu.etaxonomy.cdm.model.name.TaxonNameBase; |
| 46 | 42 |
import eu.etaxonomy.cdm.model.name.ZoologicalName; |
| 47 | 43 |
import eu.etaxonomy.cdm.model.occurrence.Collection; |
| 48 | 44 |
import eu.etaxonomy.cdm.model.occurrence.DerivationEvent; |
| 49 |
import eu.etaxonomy.cdm.model.occurrence.DerivationEventType; |
|
| 50 |
import eu.etaxonomy.cdm.model.occurrence.DerivedUnitBase; |
|
| 51 | 45 |
import eu.etaxonomy.cdm.model.occurrence.FieldObservation; |
| 52 | 46 |
import eu.etaxonomy.cdm.model.occurrence.Specimen; |
| 53 | 47 |
import eu.etaxonomy.cdm.model.reference.Reference; |
| 54 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
|
| 55 |
import eu.etaxonomy.cdm.model.reference.ReferenceType; |
|
| 56 | 48 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
| 57 | 49 |
import eu.etaxonomy.cdm.model.taxon.SynonymRelationshipType; |
| 58 | 50 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
| ... | ... | |
| 150 | 142 |
|
| 151 | 143 |
ZoologicalName name = null; |
| 152 | 144 |
if (isBlank(specSystaxRank) ){
|
| 153 |
name = makeName(state, rs); |
|
| 145 |
name = makeName(state, rs, specTaxId);
|
|
| 154 | 146 |
}else if (specSystaxRank.equals("synonym")){
|
| 155 |
Synonym synonym = getSynonym(state, rs); |
|
| 147 |
Synonym synonym = getSynonym(state, rs, specTaxId);
|
|
| 156 | 148 |
if (acceptedTaxon == null){
|
| 157 | 149 |
if (acceptedTaxonId == null){
|
| 158 | 150 |
logger.warn("Synonym has no accepted taxon defined. SpecTaxId: "+ specTaxId);
|
| ... | ... | |
| 168 | 160 |
thisTaxon = acceptedTaxon; |
| 169 | 161 |
}else{
|
| 170 | 162 |
logger.warn(String.format("Unhandled specSystaxRank %s in specTaxId %d", specSystaxRank, specTaxId));
|
| 171 |
name = makeName(state, rs); |
|
| 163 |
name = makeName(state, rs, specTaxId);
|
|
| 172 | 164 |
} |
| 173 | 165 |
|
| 174 | 166 |
if (thisTaxon != null){
|
| 175 | 167 |
name = CdmBase.deproxy(thisTaxon.getName(), ZoologicalName.class); |
| 176 |
} |
|
| 177 |
if (name == null){
|
|
| 178 |
name = makeName(state, rs); |
|
| 168 |
}else{
|
|
| 169 |
if (name == null){
|
|
| 170 |
name = makeName(state, rs, specTaxId); |
|
| 171 |
} |
|
| 172 |
thisTaxon = Taxon.NewInstance(name, sourceRef); |
|
| 173 |
objectsToSave.add(thisTaxon); |
|
| 179 | 174 |
} |
| 180 | 175 |
|
| 181 | 176 |
handleNomRef(state, referenceMap, rs, name); |
| ... | ... | |
| 214 | 209 |
|
| 215 | 210 |
|
| 216 | 211 |
private void handleTypeInformation(GlobisImportState state, ResultSet rs, ZoologicalName name, Integer specTaxId) throws SQLException {
|
| 217 |
|
|
| 218 |
|
|
| 219 | 212 |
if (! hasTypeInformation(rs)){
|
| 220 | 213 |
return; |
| 221 | 214 |
} |
| ... | ... | |
| 252 | 245 |
specTypeDepositoryStr = makeAdditionalSpecimenInformation( |
| 253 | 246 |
specTypeDepositoryStr, specimen, specTaxId); |
| 254 | 247 |
|
| 255 |
Collection collection = makeCollection(specTypeDepositoryStr, specimen, specTaxId); |
|
| 248 |
Collection collection = makeCollection(state, specTypeDepositoryStr, specimen, specTaxId);
|
|
| 256 | 249 |
String collectionCode = collection.getCode(); |
| 257 | 250 |
if (isBlank(collectionCode)){
|
| 258 | 251 |
collectionCode = collection.getName(); |
| ... | ... | |
| 302 | 295 |
|
| 303 | 296 |
|
| 304 | 297 |
/** |
| 298 |
* @param state |
|
| 305 | 299 |
* @param specTypeDepositoryStr |
| 306 | 300 |
* @param specimen |
| 307 | 301 |
* @param specTaxId |
| 308 | 302 |
*/ |
| 309 |
protected Collection makeCollection(String specTypeDepositoryStr, Specimen specimen, Integer specTaxId) {
|
|
| 310 |
//TODO deduplicate |
|
| 311 |
Map<String, Collection> collectionMap = new HashMap<String, Collection>(); |
|
| 303 |
protected Collection makeCollection(GlobisImportState state, String specTypeDepositoryStr, Specimen specimen, Integer specTaxId) {
|
|
| 312 | 304 |
|
| 313 | 305 |
//Collection |
| 314 | 306 |
specTypeDepositoryStr = specTypeDepositoryStr.replace("Washington, D.C.", "Washington@ D.C.");
|
| 315 | 307 |
|
| 316 | 308 |
Collection collection; |
| 317 | 309 |
if (specTypeDepositoryStr.equals("BMNH, London and/or MNHN, Paris")){
|
| 318 |
//TODO deduplicate |
|
| 319 |
collection = Collection.NewInstance(); |
|
| 320 |
collection.setName(specTypeDepositoryStr); |
|
| 310 |
collection = state.getRelatedObject(COLLECTION_NAMESPACE, specTypeDepositoryStr, Collection.class); |
|
| 311 |
if (collection == null){
|
|
| 312 |
collection = Collection.NewInstance(); |
|
| 313 |
collection.setName(specTypeDepositoryStr); |
|
| 314 |
collection.setTownOrLocation("London or Paris");
|
|
| 315 |
state.addRelatedObject(COLLECTION_NAMESPACE, collection.getName(), collection); |
|
| 316 |
} |
|
| 321 | 317 |
specimen.setCollection(collection); |
| 322 | 318 |
}else if (specTypeDepositoryStr.equals("coll. L. V. Kaabak, A .V. Sotshivko & V. V. Titov, Moscow")){
|
| 323 |
collection = Collection.NewInstance(); |
|
| 324 |
collection.setName("coll. L. V. Kaabak, A .V. Sotshivko & V. V. Titov");
|
|
| 325 |
collection.setTownOrLocation("Moscow");
|
|
| 319 |
String colName = "coll. L. V. Kaabak, A .V. Sotshivko & V. V. Titov"; |
|
| 320 |
collection = state.getRelatedObject(COLLECTION_NAMESPACE, colName, Collection.class); |
|
| 321 |
if (collection == null){
|
|
| 322 |
collection = Collection.NewInstance(); |
|
| 323 |
collection.setName(colName); |
|
| 324 |
collection.setTownOrLocation("Moscow");
|
|
| 325 |
state.addRelatedObject(COLLECTION_NAMESPACE, collection.getName(), collection); |
|
| 326 |
} |
|
| 326 | 327 |
specimen.setCollection(collection); |
| 327 | 328 |
}else if (specTypeDepositoryStr.matches("coll. R. E. Parrott?, Port Hope, Ontario")){
|
| 328 |
//TODO deduplicate |
|
| 329 |
collection = Collection.NewInstance(); |
|
| 330 |
collection.setName("coll. R. E. Parrott");
|
|
| 331 |
collection.setTownOrLocation("Port Hope, Ontario");
|
|
| 329 |
String colName = "coll. R. E. Parrott"; |
|
| 330 |
collection = state.getRelatedObject(COLLECTION_NAMESPACE, colName, Collection.class); |
|
| 331 |
if (collection == null){
|
|
| 332 |
collection = Collection.NewInstance(); |
|
| 333 |
collection.setName(colName); |
|
| 334 |
collection.setTownOrLocation("Port Hope, Ontario");
|
|
| 335 |
state.addRelatedObject(COLLECTION_NAMESPACE, collection.getName(), collection); |
|
| 336 |
} |
|
| 332 | 337 |
specimen.setCollection(collection); |
| 333 | 338 |
}else{
|
| 334 | 339 |
|
| 335 | 340 |
String[] split = specTypeDepositoryStr.split(",");
|
| 336 | 341 |
if (split.length != 2){
|
| 337 | 342 |
if (split.length == 1 && split[0].startsWith("coll.")){
|
| 338 |
collection = Collection.NewInstance(); |
|
| 339 |
collection.setName(split[0]); |
|
| 343 |
collection = state.getRelatedObject(COLLECTION_NAMESPACE, split[0], Collection.class); |
|
| 344 |
if (collection == null){
|
|
| 345 |
collection = Collection.NewInstance(); |
|
| 346 |
collection.setName(split[0]); |
|
| 347 |
state.addRelatedObject(COLLECTION_NAMESPACE, collection.getName(), collection); |
|
| 348 |
} |
|
| 340 | 349 |
specimen.setCollection(collection); |
| 341 | 350 |
}else{
|
| 342 | 351 |
logger.warn("Split size is not 2: " + specTypeDepositoryStr + " (specTaxID:" + specTaxId + ")");
|
| 343 | 352 |
collection = Collection.NewInstance(); |
| 344 | 353 |
collection.setCode("??");
|
| 354 |
//TODO deduplicate ?? |
|
| 345 | 355 |
} |
| 346 | 356 |
|
| 347 | 357 |
}else{
|
| 348 | 358 |
String collectionStr = split[0]; |
| 349 | 359 |
String location = split[1].replace("Washington@ D.C.", "Washington, D.C.");
|
| 350 | 360 |
|
| 351 |
collection = collectionMap.get(collectionStr);
|
|
| 361 |
collection = state.getRelatedObject(COLLECTION_NAMESPACE, collectionStr, Collection.class);
|
|
| 352 | 362 |
if (collection == null){
|
| 353 | 363 |
collection = Collection.NewInstance(); |
| 354 | 364 |
collection.setCode(collectionStr); |
| 355 | 365 |
collection.setTownOrLocation(split[1]); |
| 356 |
}else if (CdmUtils.nullSafeEqual(location, collection.getTownOrLocation())){
|
|
| 366 |
state.addRelatedObject(COLLECTION_NAMESPACE, collection.getCode(), collection); |
|
| 367 |
|
|
| 368 |
}else if (! CdmUtils.nullSafeEqual(location, collection.getTownOrLocation())){
|
|
| 357 | 369 |
String message = "Location (%s) is not equal to location (%s) of existing collection"; |
| 358 | 370 |
logger.warn(String.format(message, location, collection.getTownOrLocation(), collection.getCode())); |
| 359 | 371 |
} |
| ... | ... | |
| 540 | 552 |
|
| 541 | 553 |
|
| 542 | 554 |
|
| 543 |
private Synonym getSynonym(GlobisImportState state, ResultSet rs) throws SQLException {
|
|
| 544 |
ZoologicalName name = makeName(state, rs); |
|
| 555 |
private Synonym getSynonym(GlobisImportState state, ResultSet rs, Integer specTaxId) throws SQLException {
|
|
| 556 |
ZoologicalName name = makeName(state, rs, specTaxId);
|
|
| 545 | 557 |
|
| 546 | 558 |
Synonym synonym = Synonym.NewInstance(name, state.getTransactionalSourceReference()); |
| 547 | 559 |
|
| ... | ... | |
| 554 | 566 |
/** |
| 555 | 567 |
* @param state |
| 556 | 568 |
* @param rs |
| 569 |
* @param specTaxId |
|
| 557 | 570 |
* @return |
| 558 | 571 |
* @throws SQLException |
| 559 | 572 |
*/ |
| 560 |
protected ZoologicalName makeName(GlobisImportState state, ResultSet rs) |
|
| 573 |
protected ZoologicalName makeName(GlobisImportState state, ResultSet rs, Integer specTaxId)
|
|
| 561 | 574 |
throws SQLException {
|
| 562 | 575 |
//rank |
| 563 | 576 |
String rankStr = rs.getString("SpecRank");
|
| ... | ... | |
| 580 | 593 |
String yearStr = rs.getString("SpecYear");
|
| 581 | 594 |
String authorAndYearStr = CdmUtils.concat(", ", authorStr, yearStr);
|
| 582 | 595 |
handleAuthorAndYear(authorAndYearStr, name); |
| 596 |
|
|
| 597 |
name.addSource(String.valueOf(specTaxId), SPEC_TAX_NAMESPACE, state.getTransactionalSourceReference(), null); |
|
| 583 | 598 |
return name; |
| 584 | 599 |
} |
| 585 | 600 |
|
| ... | ... | |
| 611 | 626 |
} |
| 612 | 627 |
|
| 613 | 628 |
|
| 614 |
|
|
| 615 |
|
|
| 616 |
private boolean isInfraSpecies(GlobisImportState state, ResultSet rs, Rank rank) {
|
|
| 617 |
// TODO Auto-generated method stub |
|
| 618 |
return false; |
|
| 619 |
} |
|
| 620 |
|
|
| 621 |
|
|
| 622 |
|
|
| 623 |
|
|
| 624 |
private Reference<?> getJournal(GlobisImportState state, ResultSet rs, String refJournal) throws SQLException {
|
|
| 625 |
|
|
| 626 |
|
|
| 627 |
Reference<?> journal = ReferenceFactory.newJournal(); |
|
| 628 |
String issn = rs.getString("RefISSN");
|
|
| 629 |
if (StringUtils.isNotBlank(issn)){
|
|
| 630 |
issn.replaceAll("ISSN", "").trim();
|
|
| 631 |
journal.setIssn(issn); |
|
| 632 |
} |
|
| 633 |
|
|
| 634 |
|
|
| 635 |
|
|
| 636 |
//TODO deduplicate |
|
| 637 |
journal.setTitle(refJournal); |
|
| 638 |
return journal; |
|
| 639 |
} |
|
| 640 |
|
|
| 641 |
|
|
| 642 |
|
|
| 643 |
|
|
| 644 |
/* (non-Javadoc) |
|
| 645 |
* @see eu.etaxonomy.cdm.io.common.mapping.IMappingImport#createObject(java.sql.ResultSet, eu.etaxonomy.cdm.io.common.ImportStateBase) |
|
| 646 |
*/ |
|
| 647 |
public Reference<?> createObject(ResultSet rs, GlobisImportState state) |
|
| 648 |
throws SQLException {
|
|
| 649 |
Reference<?> ref; |
|
| 650 |
String refType = rs.getString("RefType");
|
|
| 651 |
if (refType == null){
|
|
| 652 |
ref = ReferenceFactory.newGeneric(); |
|
| 653 |
}else if (refType == "book"){
|
|
| 654 |
ref = ReferenceFactory.newBook(); |
|
| 655 |
}else if (refType == "paper in journal"){
|
|
| 656 |
ref = ReferenceFactory.newArticle(); |
|
| 657 |
}else if (refType.startsWith("unpublished") ){
|
|
| 658 |
ref = ReferenceFactory.newGeneric(); |
|
| 659 |
}else if (refType.endsWith("paper in journal")){
|
|
| 660 |
ref = ReferenceFactory.newArticle(); |
|
| 661 |
}else if (refType == "paper in book"){
|
|
| 662 |
ref = ReferenceFactory.newBookSection(); |
|
| 663 |
}else if (refType == "paper in journalwebsite"){
|
|
| 664 |
ref = ReferenceFactory.newArticle(); |
|
| 665 |
}else{
|
|
| 666 |
logger.warn("Unknown reference type: " + refType);
|
|
| 667 |
ref = ReferenceFactory.newGeneric(); |
|
| 668 |
} |
|
| 669 |
return ref; |
|
| 670 |
} |
|
| 671 |
|
|
| 672 | 629 |
/* (non-Javadoc) |
| 673 | 630 |
* @see eu.etaxonomy.cdm.io.berlinModel.in.IPartitionedIO#getRelatedObjectsForPartition(java.sql.ResultSet) |
| 674 | 631 |
*/ |
| ... | ... | |
| 700 | 657 |
Map<String, Reference> referenceMap = (Map<String, Reference>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace); |
| 701 | 658 |
result.put(nameSpace, referenceMap); |
| 702 | 659 |
|
| 660 |
//collection map |
|
| 661 |
nameSpace = COLLECTION_NAMESPACE; |
|
| 662 |
List<Collection> listCollection = getCollectionService().list(Collection.class, null, null, null, null); |
|
| 663 |
Map<String, Collection> collectionMap = new HashMap<String, Collection>(); |
|
| 664 |
for (Collection collection : listCollection){
|
|
| 665 |
collectionMap.put(collection.getCode(), collection); |
|
| 666 |
if (isBlank(collection.getCode())){
|
|
| 667 |
logger.warn("Collection code is blank: " + collection);
|
|
| 668 |
} |
|
| 669 |
} |
|
| 670 |
result.put(nameSpace, collectionMap); |
|
| 703 | 671 |
|
| 704 | 672 |
} catch (SQLException e) {
|
| 705 | 673 |
throw new RuntimeException(e); |
| ... | ... | |
| 727 | 695 |
|
| 728 | 696 |
|
| 729 | 697 |
|
| 698 |
@Override |
|
| 699 |
public Reference createObject(ResultSet rs, GlobisImportState state) |
|
| 700 |
throws SQLException {
|
|
| 701 |
// not needed |
|
| 702 |
return null; |
|
| 703 |
} |
|
| 704 |
|
|
| 705 |
|
|
| 706 |
|
|
| 707 |
|
|
| 730 | 708 |
|
| 731 | 709 |
} |
Also available in: Unified diff
latest globis