Revision bb38665e
Added by Andreas Müller over 10 years ago
| app-import/src/main/java/eu/etaxonomy/cdm/io/algaterra/AlgaTerraDnaImport.java | ||
|---|---|---|
| 27 | 27 |
import org.springframework.format.datetime.joda.DateTimeParser; |
| 28 | 28 |
import org.springframework.stereotype.Component; |
| 29 | 29 |
|
| 30 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade.DerivedUnitType; |
|
| 31 | 30 |
import eu.etaxonomy.cdm.io.algaterra.validation.AlgaTerraDnaImportValidator; |
| 32 | 31 |
import eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportConfigurator; |
| 33 | 32 |
import eu.etaxonomy.cdm.io.berlinModel.in.BerlinModelImportState; |
| ... | ... | |
| 37 | 36 |
import eu.etaxonomy.cdm.model.common.Annotation; |
| 38 | 37 |
import eu.etaxonomy.cdm.model.common.AnnotationType; |
| 39 | 38 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
| 39 |
import eu.etaxonomy.cdm.model.common.DefinedTerm; |
|
| 40 | 40 |
import eu.etaxonomy.cdm.model.common.Language; |
| 41 | 41 |
import eu.etaxonomy.cdm.model.description.IndividualsAssociation; |
| 42 | 42 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
| 43 | 43 |
import eu.etaxonomy.cdm.model.molecular.DnaSample; |
| 44 |
import eu.etaxonomy.cdm.model.molecular.GenBankAccession; |
|
| 45 |
import eu.etaxonomy.cdm.model.molecular.Locus; |
|
| 46 | 44 |
import eu.etaxonomy.cdm.model.molecular.Sequence; |
| 47 | 45 |
import eu.etaxonomy.cdm.model.occurrence.DerivationEvent; |
| 48 | 46 |
import eu.etaxonomy.cdm.model.occurrence.DerivationEventType; |
| 49 |
import eu.etaxonomy.cdm.model.occurrence.DerivedUnitBase;
|
|
| 50 |
import eu.etaxonomy.cdm.model.occurrence.FieldObservation;
|
|
| 47 |
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit; |
|
| 48 |
import eu.etaxonomy.cdm.model.occurrence.FieldUnit;
|
|
| 51 | 49 |
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase; |
| 52 | 50 |
import eu.etaxonomy.cdm.model.reference.Reference; |
| 53 | 51 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
| ... | ... | |
| 124 | 122 |
Set<SpecimenOrObservationBase> samplesToSave = new HashSet<SpecimenOrObservationBase>(); |
| 125 | 123 |
Set<TaxonBase> taxaToSave = new HashSet<TaxonBase>(); |
| 126 | 124 |
|
| 127 |
Map<String, FieldObservation> ecoFactFieldObservationMap = (Map<String, FieldObservation>) partitioner.getObjectMap(ECO_FACT_FIELD_OBSERVATION_NAMESPACE);
|
|
| 125 |
Map<String, FieldUnit> ecoFactFieldObservationMap = (Map<String, FieldUnit>) partitioner.getObjectMap(ECO_FACT_FIELD_OBSERVATION_NAMESPACE);
|
|
| 128 | 126 |
|
| 129 | 127 |
ResultSet rs = partitioner.getResultSet(); |
| 130 | 128 |
|
| ... | ... | |
| 165 | 163 |
Sequence sequence = makeSequence(rs, dnaSample, dnaFactId, importDateTime); |
| 166 | 164 |
|
| 167 | 165 |
//locus |
| 168 |
Locus locus = Locus.NewInstance(keywordsStr, definitionStr); |
|
| 166 |
//FIXME Deduplicate DnaMarker |
|
| 167 |
DefinedTerm locus = DefinedTerm.NewDnaMarkerInstance(definitionStr, keywordsStr, null); |
|
| 169 | 168 |
locus.setCreated(importDateTime); |
| 170 |
sequence.setLocus(locus); |
|
| 169 |
this.getTermService().save(locus); |
|
| 170 |
|
|
| 171 |
sequence.setDnaMarker(locus); |
|
| 171 | 172 |
|
| 172 | 173 |
//GenBank Accession |
| 173 | 174 |
makeGenBankAccession(rs, sequence, importDateTime, dnaFactId); |
| ... | ... | |
| 198 | 199 |
ref.setTitleCache(referenceStr, true); |
| 199 | 200 |
referenceMap.put(referenceStr, ref); |
| 200 | 201 |
} |
| 201 |
sequence.setPublishedIn(ref);
|
|
| 202 |
sequence.addCitation(ref);
|
|
| 202 | 203 |
} |
| 203 | 204 |
|
| 204 | 205 |
//save |
| ... | ... | |
| 229 | 230 |
Integer ecoFactFk = nullSafeInt(rs, "ecoFactId"); |
| 230 | 231 |
if (ecoFactFk != null){
|
| 231 | 232 |
|
| 232 |
DerivedUnitBase<?> ecoFact = (DerivedUnitBase<?>)state.getRelatedObject(ECO_FACT_DERIVED_UNIT_NAMESPACE, ecoFactFk.toString());
|
|
| 233 |
DerivedUnit ecoFact = (DerivedUnit)state.getRelatedObject(ECO_FACT_DERIVED_UNIT_NAMESPACE, ecoFactFk.toString());
|
|
| 233 | 234 |
if (ecoFact == null){
|
| 234 | 235 |
logger.warn("EcoFact is null for ecoFactFk: " + ecoFactFk + ", DnaFactId: " + dnaFactId);
|
| 235 | 236 |
}else{
|
| ... | ... | |
| 284 | 285 |
|
| 285 | 286 |
private Sequence makeSequence(ResultSet rs, DnaSample dnaSample, int dnaFactId, DateTime importDateTime) throws SQLException {
|
| 286 | 287 |
String sequenceStr = rs.getString("PlainSequence");
|
| 287 |
Integer originalLen = null; |
|
| 288 | 288 |
Integer seqLen = nullSafeInt(rs, "SeqLen"); |
| 289 |
if (seqLen == null){
|
|
| 290 |
if (sequenceStr != null){
|
|
| 291 |
seqLen = sequenceStr.length(); |
|
| 292 |
} |
|
| 293 |
} |
|
| 294 |
|
|
| 295 |
if (sequenceStr != null){
|
|
| 296 |
originalLen = sequenceStr.length(); |
|
| 297 |
if (originalLen > 255){
|
|
| 298 |
logger.warn("Sequence truncated. Id: " + dnaFactId);
|
|
| 299 |
sequenceStr = sequenceStr.substring(0, 255); |
|
| 300 |
} |
|
| 301 |
}else{
|
|
| 289 |
|
|
| 290 |
if (sequenceStr == null){
|
|
| 302 | 291 |
logger.warn("PlainSequence is null. Id: " + dnaFactId);
|
| 303 |
} |
|
| 304 |
Sequence sequence = Sequence.NewInstance(sequenceStr); |
|
| 305 |
sequence.setLength(seqLen); |
|
| 306 |
if (! originalLen.equals(seqLen)){
|
|
| 307 |
logger.warn("SeqLen (" + seqLen+ ") and OriginalLen ("+originalLen+") differ for dnaFact: " + dnaFactId);
|
|
| 292 |
}else{
|
|
| 293 |
if (sequenceStr.length() != seqLen){
|
|
| 294 |
logger.warn("SeqLen (" + seqLen+ ") and OriginalLen ("+sequenceStr.length()+") differ for dnaFact: " + dnaFactId);
|
|
| 295 |
} |
|
| 308 | 296 |
} |
| 309 | 297 |
|
| 298 |
Sequence sequence = Sequence.NewInstance(sequenceStr, seqLen); |
|
| 310 | 299 |
sequence.setCreated(importDateTime); |
| 311 |
dnaSample.addSequences(sequence);
|
|
| 300 |
dnaSample.addSequence(sequence); |
|
| 312 | 301 |
return sequence; |
| 313 | 302 |
} |
| 314 | 303 |
|
| ... | ... | |
| 343 | 332 |
accessionStr = null; |
| 344 | 333 |
} |
| 345 | 334 |
if (isGenBankAccessionNumber(accessionStr, versionStr, genBankUri, dnaFactId) || genBankUri != null){
|
| 346 |
GenBankAccession accession = GenBankAccession.NewInstance(accessionStr); |
|
| 347 |
accession.setUri(genBankUri); |
|
| 348 |
accession.setCreated(importDateTime); |
|
| 349 |
sequence.addGenBankAccession(accession); |
|
| 335 |
sequence.setGeneticAccessionNumber(accessionStr); |
|
| 350 | 336 |
} |
| 351 | 337 |
} |
| 352 | 338 |
} |
| ... | ... | |
| 377 | 363 |
return ECO_FACT_FIELD_OBSERVATION_NAMESPACE; |
| 378 | 364 |
} |
| 379 | 365 |
|
| 380 |
|
|
| 381 |
private DerivedUnitType makeDerivedUnitType(String recordBasis) {
|
|
| 382 |
DerivedUnitType result = null; |
|
| 383 |
if (StringUtils.isBlank(recordBasis)){
|
|
| 384 |
result = DerivedUnitType.DerivedUnit; |
|
| 385 |
} else if (recordBasis.equalsIgnoreCase("FossileSpecimen")){
|
|
| 386 |
result = DerivedUnitType.Fossil; |
|
| 387 |
}else if (recordBasis.equalsIgnoreCase("HumanObservation")){
|
|
| 388 |
result = DerivedUnitType.Observation; |
|
| 389 |
}else if (recordBasis.equalsIgnoreCase("Literature")){
|
|
| 390 |
logger.warn("Literature record basis not yet supported");
|
|
| 391 |
result = DerivedUnitType.DerivedUnit; |
|
| 392 |
}else if (recordBasis.equalsIgnoreCase("LivingSpecimen")){
|
|
| 393 |
result = DerivedUnitType.LivingBeing; |
|
| 394 |
}else if (recordBasis.equalsIgnoreCase("MachineObservation")){
|
|
| 395 |
logger.warn("MachineObservation record basis not yet supported");
|
|
| 396 |
result = DerivedUnitType.Observation; |
|
| 397 |
}else if (recordBasis.equalsIgnoreCase("PreservedSpecimen")){
|
|
| 398 |
result = DerivedUnitType.Specimen; |
|
| 399 |
} |
|
| 400 |
return result; |
|
| 401 |
} |
|
| 402 |
|
|
| 403 |
/* (non-Javadoc) |
|
| 404 |
* @see eu.etaxonomy.cdm.io.berlinModel.in.IPartitionedIO#getRelatedObjectsForPartition(java.sql.ResultSet) |
|
| 405 |
*/ |
|
| 366 |
@Override |
|
| 406 | 367 |
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs) {
|
| 407 | 368 |
String nameSpace; |
| 408 |
Class cdmClass; |
|
| 369 |
Class<?> cdmClass;
|
|
| 409 | 370 |
Set<String> idSet; |
| 410 | 371 |
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>(); |
| 411 | 372 |
|
| ... | ... | |
| 430 | 391 |
|
| 431 | 392 |
//eco fact derived unit map |
| 432 | 393 |
nameSpace = AlgaTerraFactEcologyImport.ECO_FACT_DERIVED_UNIT_NAMESPACE; |
| 433 |
cdmClass = DerivedUnitBase.class;
|
|
| 394 |
cdmClass = DerivedUnit.class; |
|
| 434 | 395 |
idSet = ecoFactFkSet; |
| 435 |
Map<String, DerivedUnitBase> derivedUnitMap = (Map<String, DerivedUnitBase>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
|
| 396 |
Map<String, DerivedUnit> derivedUnitMap = (Map<String, DerivedUnit>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
|
| 436 | 397 |
result.put(nameSpace, derivedUnitMap); |
| 437 | 398 |
|
| 438 | 399 |
} catch (SQLException e) {
|
| ... | ... | |
| 441 | 402 |
return result; |
| 442 | 403 |
} |
| 443 | 404 |
|
| 444 |
|
|
| 445 |
|
|
| 446 |
/* (non-Javadoc) |
|
| 447 |
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#doCheck(eu.etaxonomy.cdm.io.common.IoStateBase) |
|
| 448 |
*/ |
|
| 449 | 405 |
@Override |
| 450 | 406 |
protected boolean doCheck(BerlinModelImportState state){
|
| 451 | 407 |
IOValidator<BerlinModelImportState> validator = new AlgaTerraDnaImportValidator(); |
Also available in: Unified diff
reintegrate cdmlib-app-3.3 branch into trunk