Revision 5ab3e456
Added by Katja Luther over 8 years ago
| cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/tcsrdf/TcsRdfTaxonNameImport.java | ||
|---|---|---|
| 9 | 9 |
|
| 10 | 10 |
package eu.etaxonomy.cdm.io.tcsrdf; |
| 11 | 11 |
|
| 12 |
import java.io.IOException; |
|
| 12 | 13 |
import java.io.InputStream; |
| 14 |
import java.util.List; |
|
| 13 | 15 |
import java.util.Set; |
| 14 | 16 |
|
| 15 | 17 |
import org.apache.log4j.Logger; |
| ... | ... | |
| 20 | 22 |
|
| 21 | 23 |
import com.hp.hpl.jena.rdf.model.Model; |
| 22 | 24 |
import com.hp.hpl.jena.rdf.model.ModelFactory; |
| 23 |
import com.hp.hpl.jena.rdf.model.Property; |
|
| 24 | 25 |
import com.hp.hpl.jena.rdf.model.RDFNode; |
| 25 | 26 |
import com.hp.hpl.jena.rdf.model.ResIterator; |
| 26 | 27 |
import com.hp.hpl.jena.rdf.model.Resource; |
| 27 | 28 |
import com.hp.hpl.jena.rdf.model.Statement; |
| 28 | 29 |
import com.hp.hpl.jena.rdf.model.StmtIterator; |
| 30 |
import com.hp.hpl.jena.rdf.model.Property; |
|
| 31 |
|
|
| 32 |
|
|
| 33 |
|
|
| 29 | 34 |
|
| 30 | 35 |
import eu.etaxonomy.cdm.common.XmlHelp; |
| 31 | 36 |
import eu.etaxonomy.cdm.io.common.ICdmIO; |
| ... | ... | |
| 44 | 49 |
import eu.etaxonomy.cdm.model.reference.Reference; |
| 45 | 50 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
| 46 | 51 |
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException; |
| 47 |
import eu.etaxonomy.cdm.strategy.parser.INonViralNameParser; |
|
| 48 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl; |
|
| 49 | 52 |
|
| 50 | 53 |
/** |
| 51 | 54 |
* @author a.mueller |
| ... | ... | |
| 57 | 60 |
private static final Logger logger = Logger.getLogger(TcsRdfTaxonNameImport.class); |
| 58 | 61 |
|
| 59 | 62 |
private static int modCount = 5000; |
| 60 |
INonViralNameParser nameParser = new NonViralNameParserImpl(); |
|
| 61 | 63 |
|
| 62 | 64 |
public TcsRdfTaxonNameImport(){
|
| 63 | 65 |
super(); |
| ... | ... | |
| 125 | 127 |
Resource elTaxonName = root.getResource(taxonNameNamespace); |
| 126 | 128 |
|
| 127 | 129 |
int i = 0; |
| 128 |
|
|
| 130 |
|
|
| 129 | 131 |
TaxonNameBase name; |
| 130 | 132 |
Property property = root.getProperty(taxonNameNamespace+"authorship"); |
| 131 |
|
|
| 133 |
|
|
| 132 | 134 |
ResIterator iterator = root.listSubjectsWithProperty(property, (RDFNode) null); |
| 133 | 135 |
String id ; |
| 134 | 136 |
while (iterator.hasNext()){
|
| 135 |
|
|
| 137 |
|
|
| 136 | 138 |
Resource resource = iterator.next(); |
| 137 |
|
|
| 139 |
|
|
| 138 | 140 |
name = handleNameResource(resource, config); |
| 139 | 141 |
id = resource.getNameSpace(); |
| 140 | 142 |
taxonNameMap.put(id, name); |
| 141 | 143 |
} |
| 142 |
|
|
| 144 |
|
|
| 143 | 145 |
logger.info(i + " names handled"); |
| 144 | 146 |
getNameService().save(taxonNameMap.objects()); |
| 145 | 147 |
// makeNameSpecificData(nameMap); |
| ... | ... | |
| 154 | 156 |
protected boolean isIgnore(TcsRdfImportState state){
|
| 155 | 157 |
return ! state.getConfig().isDoTaxonNames(); |
| 156 | 158 |
} |
| 157 |
|
|
| 159 |
|
|
| 158 | 160 |
protected TaxonNameBase handleNameModel(Model model, TcsRdfImportConfigurator config, MapWrapper<TaxonNameBase> taxonNameMap, String uri){
|
| 159 | 161 |
Resource nameAbout = model.getResource(uri); |
| 160 | 162 |
TaxonNameBase result = handleNameResource(nameAbout, config); |
| 161 | 163 |
taxonNameMap.put(uri, result); |
| 162 | 164 |
return result; |
| 163 |
|
|
| 165 |
|
|
| 164 | 166 |
} |
| 165 |
|
|
| 167 |
|
|
| 166 | 168 |
private TaxonNameBase handleNameResource(Resource nameAbout, TcsRdfImportConfigurator config){
|
| 167 | 169 |
String idNamespace = "TaxonName"; |
| 168 |
|
|
| 170 |
|
|
| 169 | 171 |
StmtIterator stmts = nameAbout.listProperties(); |
| 170 | 172 |
while(stmts.hasNext()){
|
| 171 | 173 |
System.out.println(stmts.next().getPredicate().toString()); |
| 172 | 174 |
} |
| 173 |
|
|
| 175 |
|
|
| 174 | 176 |
Property prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"nomenclaturalCode"); |
| 175 | 177 |
Statement stateNomenclaturalCode = nameAbout.getProperty(prop); |
| 176 | 178 |
String strNomenclaturalCode = stateNomenclaturalCode.getObject().toString(); |
| 177 | 179 |
//Rank |
| 178 | 180 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"rankString"); |
| 179 | 181 |
String strRank = nameAbout.getProperty(prop).getString(); |
| 180 |
|
|
| 181 |
|
|
| 182 |
|
|
| 182 |
|
|
| 183 |
|
|
| 184 |
|
|
| 183 | 185 |
try {
|
| 184 |
|
|
| 186 |
|
|
| 185 | 187 |
Rank rank = TcsRdfTransformer.rankString2Rank(strRank); |
| 186 | 188 |
NomenclaturalCode nomCode; |
| 187 | 189 |
if (strNomenclaturalCode != null){
|
| ... | ... | |
| 189 | 191 |
}else{
|
| 190 | 192 |
nomCode = NomenclaturalCode.ICNAFP; |
| 191 | 193 |
} |
| 192 |
|
|
| 194 |
|
|
| 193 | 195 |
TaxonNameBase<?,?> nameBase = nomCode.getNewTaxonNameInstance(rank); |
| 194 | 196 |
|
| 195 | 197 |
Set<String> omitAttributes = null; |
| 196 | 198 |
//makeStandardMapper(nameAbout, nameBase, omitAttributes, standardMappers); |
| 197 |
|
|
| 199 |
|
|
| 198 | 200 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"nameComplete"); |
| 199 | 201 |
String strNameComplete = nameAbout.getProperty(prop).getString(); |
| 200 | 202 |
nameBase.setTitleCache(strNameComplete, true); |
| 201 |
|
|
| 203 |
|
|
| 202 | 204 |
prop = nameAbout.getModel().getProperty(config.getCommonNamespaceURIString()+"publishedIn"); |
| 203 | 205 |
String strPublishedIn = nameAbout.getProperty(prop).getString(); |
| 204 | 206 |
if (strPublishedIn != null && strPublishedIn != ""){
|
| ... | ... | |
| 224 | 226 |
((Reference<?>)nomRef).addMarker(Marker.NewInstance(MarkerType.PUBLISH(), false)); |
| 225 | 227 |
} |
| 226 | 228 |
} |
| 227 |
|
|
| 229 |
|
|
| 228 | 230 |
if (nameBase instanceof NonViralName){
|
| 229 | 231 |
NonViralName<?> nonViralName = (NonViralName<?>)nameBase; |
| 230 | 232 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"genusPart"); |
| ... | ... | |
| 235 | 237 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"uninomial"); |
| 236 | 238 |
strGenusPart = nameAbout.getProperty(prop).getString(); |
| 237 | 239 |
} |
| 238 |
|
|
| 240 |
|
|
| 239 | 241 |
nonViralName.setGenusOrUninomial(strGenusPart); |
| 240 |
|
|
| 242 |
|
|
| 241 | 243 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"infragenericEpithet"); |
| 242 | 244 |
try{
|
| 243 | 245 |
String strInfragenericEpithet = nameAbout.getProperty(prop).getString(); |
| 244 | 246 |
nonViralName.setInfraGenericEpithet(strInfragenericEpithet); |
| 245 | 247 |
}catch(NullPointerException e){
|
| 246 |
|
|
| 248 |
|
|
| 247 | 249 |
} |
| 248 | 250 |
try {
|
| 249 | 251 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"specificEpithet"); |
| 250 | 252 |
String strSpecificEpithet = nameAbout.getProperty(prop).getString(); |
| 251 | 253 |
nonViralName.setSpecificEpithet(strSpecificEpithet); |
| 252 | 254 |
}catch(NullPointerException e){
|
| 253 |
|
|
| 255 |
|
|
| 254 | 256 |
} |
| 255 | 257 |
try{
|
| 256 | 258 |
prop = nameAbout.getModel().getProperty(config.getTnNamespaceURIString()+"infraspecificEpithet"); |
| 257 | 259 |
String strInfraspecificEpithet = nameAbout.getProperty(prop).getString(); |
| 258 | 260 |
nonViralName.setInfraSpecificEpithet(strInfraspecificEpithet); |
| 259 | 261 |
}catch(NullPointerException e){
|
| 260 |
|
|
| 262 |
|
|
| 261 | 263 |
} |
| 262 | 264 |
//Authorships |
| 263 | 265 |
//TODO |
| ... | ... | |
| 268 | 270 |
<tm:hasMember rdf:resource="urn:lsid:ipni.org:authors:2691-1" |
| 269 | 271 |
tm:index="1" |
| 270 | 272 |
tm:role="Combination Author"/> |
| 271 |
<tm:hasMember rdf:resource="urn:lsid:ipni.org:authors:8096-1" |
|
| 273 |
<tm:hasMember rdf:resource="urn:lsid:ipni.org:authors:8096-1"
|
|
| 272 | 274 |
tm:index="1" |
| 273 | 275 |
tm:role="Basionym Author"/> |
| 274 | 276 |
</tm:Team> |
| ... | ... | |
| 288 | 290 |
try{
|
| 289 | 291 |
stateAutorTeamTeam = stateAuthorTeam.getProperty(prop); |
| 290 | 292 |
}catch(Exception e){
|
| 291 |
|
|
| 293 |
|
|
| 292 | 294 |
} |
| 293 | 295 |
try{
|
| 294 | 296 |
prop = stateAuthorTeam.getModel().getProperty(config.getTeamNamespaceURIString()+"name"); |
| 295 | 297 |
stateAutorTeamName = stateAuthorTeam.getProperty(prop); |
| 296 | 298 |
}catch(Exception e){
|
| 297 |
|
|
| 299 |
|
|
| 298 | 300 |
} |
| 299 | 301 |
try{
|
| 300 | 302 |
prop = nameAbout.getModel().getProperty(config.getTeamNamespaceURIString()+"hasMember"); |
| 301 | 303 |
stateTeamMember = ((Resource)stateAuthorTeam.getObject()).listProperties(prop); |
| 302 | 304 |
String memberString = null; |
| 303 | 305 |
Person person; |
| 304 |
if (stateTeamMember.toList().size() ==1){
|
|
| 305 |
person = Person.NewTitledInstance(authorTeam.getTitleCache()); |
|
| 306 |
}else {
|
|
| 307 |
nameParser.parseAuthors(nonViralName, authorTeam.getTitleCache()); |
|
| 308 |
} |
|
| 309 |
/* for (Statement statement :stateTeamMember.toList()){
|
|
| 306 |
for (Statement statement :stateTeamMember.toList()){
|
|
| 310 | 307 |
memberString =statement.getObject().toString(); |
| 311 | 308 |
if (memberString != null){
|
| 312 | 309 |
person = Person.NewTitledInstance(memberString); |
| 313 | 310 |
authorTeam.addTeamMember(person); |
| 314 | 311 |
} |
| 315 |
}*/
|
|
| 312 |
} |
|
| 316 | 313 |
}catch(Exception e){
|
| 317 | 314 |
System.err.println(e.getMessage()); |
| 318 | 315 |
} |
| 319 | 316 |
} |
| 320 |
|
|
| 321 |
|
|
| 322 |
|
|
| 317 |
|
|
| 318 |
|
|
| 319 |
|
|
| 323 | 320 |
nonViralName.setCombinationAuthorship(authorTeam); |
| 324 |
|
|
| 321 |
|
|
| 325 | 322 |
//Annotations: |
| 326 | 323 |
/* |
| 327 | 324 |
* <tn:hasAnnotation> |
| ... | ... | |
| 329 | 326 |
<tn:noteType rdf:resource="http://rs.tdwg.org/ontology/voc/TaxonName#replacementNameFor"/> |
| 330 | 327 |
<tn:objectTaxonName rdf:resource="urn:lsid:ipni.org:names:151538-1"/> |
| 331 | 328 |
</tn:NomenclaturalNote> |
| 332 |
</tn:hasAnnotation> |
|
| 329 |
</tn:hasAnnotation>
|
|
| 333 | 330 |
*/ |
| 334 | 331 |
/* |
| 335 | 332 |
String strInfraspecificEpithet = nameAbout.getProperty(prop).getString(); |
| ... | ... | |
| 373 | 370 |
//ImportHelper.setOriginalSource(nameBase, config.getSourceReference(), nameAbout, idNamespace); |
| 374 | 371 |
|
| 375 | 372 |
//checkAdditionalContents(elTaxonName, standardMappers, operationalMappers, unclearMappers); |
| 376 |
|
|
| 373 |
|
|
| 377 | 374 |
return nameBase; |
| 378 |
|
|
| 375 |
|
|
| 379 | 376 |
}catch(Exception e){
|
| 380 | 377 |
e.printStackTrace(); |
| 381 | 378 |
return null; |
| ... | ... | |
| 384 | 381 |
//name |
| 385 | 382 |
String strNameComplete = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "nameComplete", rdfNamespace); |
| 386 | 383 |
nameBase.setTitleCache(strNameComplete, true); |
| 387 |
|
|
| 384 |
|
|
| 388 | 385 |
//Reference |
| 389 | 386 |
//TODO |
| 390 | 387 |
String tcsElementName = "publishedIn"; |
| ... | ... | |
| 442 | 439 |
if (nameBase instanceof NonViralName){
|
| 443 | 440 |
NonViralName<?> nonViralName = (NonViralName<?>)nameBase; |
| 444 | 441 |
String strGenusPart = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "genusPart", rdfNamespace); |
| 445 |
|
|
| 442 |
|
|
| 446 | 443 |
//for names of rank genus the uninomial property should be used |
| 447 | 444 |
if (strGenusPart == null){
|
| 448 | 445 |
strGenusPart = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "uninomial", rdfNamespace); |
| 449 | 446 |
} |
| 450 | 447 |
nonViralName.setGenusOrUninomial(strGenusPart); |
| 451 |
|
|
| 448 |
|
|
| 452 | 449 |
String strInfragenericEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "infragenericEpithet", rdfNamespace); |
| 453 | 450 |
nonViralName.setGenusOrUninomial(strInfragenericEpithet); |
| 454 |
|
|
| 455 |
|
|
| 456 |
|
|
| 451 |
|
|
| 452 |
|
|
| 453 |
|
|
| 457 | 454 |
String strSpecificEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "specificEpithet", rdfNamespace); |
| 458 | 455 |
nonViralName.setSpecificEpithet(strSpecificEpithet); |
| 459 |
|
|
| 456 |
|
|
| 460 | 457 |
String strInfraspecificEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "infraspecificEpithet", rdfNamespace); |
| 461 | 458 |
nonViralName.setInfraSpecificEpithet(strInfraspecificEpithet); |
| 462 | 459 |
//AuthorTeams |
| ... | ... | |
| 512 | 509 |
} |
| 513 | 510 |
return null;*/ |
| 514 | 511 |
} |
| 515 |
|
|
| 512 |
|
|
| 516 | 513 |
protected TaxonNameBase handleNameElement(Element elTaxonName, Namespace rdfNamespace, Namespace taxonNameNamespace, TcsRdfImportConfigurator config, MapWrapper<TaxonNameBase> taxonNameMap){
|
| 517 | 514 |
String idNamespace = "TaxonName"; |
| 518 | 515 |
Attribute about = elTaxonName.getAttribute("about", rdfNamespace);
|
| 519 |
|
|
| 516 |
|
|
| 520 | 517 |
//create TaxonName element |
| 521 |
|
|
| 522 |
|
|
| 518 |
|
|
| 519 |
|
|
| 523 | 520 |
String nameAbout = elTaxonName.getAttributeValue("about", rdfNamespace);
|
| 524 | 521 |
if (nameAbout == null){
|
| 525 | 522 |
nameAbout = XmlHelp.getChildAttributeValue(elTaxonName, "TaxonName", taxonNameNamespace, "about", rdfNamespace); |
| 526 | 523 |
} |
| 527 |
|
|
| 528 |
|
|
| 524 |
|
|
| 525 |
|
|
| 529 | 526 |
String strRank = XmlHelp.getChildAttributeValue(elTaxonName, "rankString", taxonNameNamespace, "rankString", rdfNamespace); |
| 530 | 527 |
if (strRank == null){
|
| 531 | 528 |
strRank = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "rankString", rdfNamespace); |
| 532 |
|
|
| 529 |
|
|
| 533 | 530 |
} |
| 534 |
|
|
| 531 |
|
|
| 535 | 532 |
if (strRank == null){
|
| 536 | 533 |
strRank = XmlHelp.getChildAttributeValue(elTaxonName, "rank", taxonNameNamespace, "resource", rdfNamespace); |
| 537 |
|
|
| 534 |
|
|
| 538 | 535 |
} |
| 539 |
|
|
| 536 |
|
|
| 540 | 537 |
String strNomenclaturalCode = XmlHelp.getChildContentAttributeValue(elTaxonName, "TaxonName", taxonNameNamespace, "nomenclaturalCode", rdfNamespace); |
| 541 | 538 |
if (strNomenclaturalCode == null){
|
| 542 | 539 |
strNomenclaturalCode = XmlHelp.getChildAttributeValue(elTaxonName, "nomenclaturalCode", taxonNameNamespace, "resource", rdfNamespace); |
| 543 |
|
|
| 540 |
|
|
| 544 | 541 |
} |
| 545 | 542 |
try {
|
| 546 |
|
|
| 543 |
|
|
| 547 | 544 |
Rank rank = TcsRdfTransformer.rankString2Rank(strRank); |
| 548 | 545 |
NomenclaturalCode nomCode; |
| 549 | 546 |
if (strNomenclaturalCode != null){
|
| ... | ... | |
| 551 | 548 |
}else{
|
| 552 | 549 |
nomCode = NomenclaturalCode.ICNAFP; |
| 553 | 550 |
} |
| 554 |
|
|
| 551 |
|
|
| 555 | 552 |
TaxonNameBase<?,?> nameBase = nomCode.getNewTaxonNameInstance(rank); |
| 556 | 553 |
|
| 557 | 554 |
Set<String> omitAttributes = null; |
| 558 | 555 |
//makeStandardMapper(elTaxonName, nameBase, omitAttributes, standardMappers); |
| 559 |
|
|
| 556 |
|
|
| 560 | 557 |
//name |
| 561 | 558 |
String strNameComplete = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "nameComplete", rdfNamespace); |
| 562 | 559 |
nameBase.setTitleCache(strNameComplete, true); |
| 563 |
|
|
| 560 |
|
|
| 564 | 561 |
//Reference |
| 565 | 562 |
//TODO |
| 566 | 563 |
String tcsElementName = "publishedIn"; |
| ... | ... | |
| 618 | 615 |
if (nameBase instanceof NonViralName){
|
| 619 | 616 |
NonViralName<?> nonViralName = (NonViralName<?>)nameBase; |
| 620 | 617 |
String strGenusPart = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "genusPart", rdfNamespace); |
| 621 |
|
|
| 618 |
|
|
| 622 | 619 |
//for names of rank genus the uninomial property should be used |
| 623 | 620 |
if (strGenusPart == null){
|
| 624 | 621 |
strGenusPart = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "uninomial", rdfNamespace); |
| 625 | 622 |
} |
| 626 | 623 |
nonViralName.setGenusOrUninomial(strGenusPart); |
| 627 |
|
|
| 624 |
|
|
| 628 | 625 |
String strInfragenericEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "infragenericEpithet", rdfNamespace); |
| 629 | 626 |
nonViralName.setGenusOrUninomial(strInfragenericEpithet); |
| 630 |
|
|
| 631 |
|
|
| 632 |
|
|
| 627 |
|
|
| 628 |
|
|
| 629 |
|
|
| 633 | 630 |
String strSpecificEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "specificEpithet", rdfNamespace); |
| 634 | 631 |
nonViralName.setSpecificEpithet(strSpecificEpithet); |
| 635 |
|
|
| 632 |
|
|
| 636 | 633 |
String strInfraspecificEpithet = XmlHelp.getChildContent(elTaxonName, "TaxonName", taxonNameNamespace, "infraspecificEpithet", rdfNamespace); |
| 637 | 634 |
nonViralName.setInfraSpecificEpithet(strInfraspecificEpithet); |
| 638 | 635 |
//AuthorTeams |
| ... | ... | |
| 693 | 690 |
Model model = ModelFactory.createDefaultModel(); |
| 694 | 691 |
try{
|
| 695 | 692 |
model.read(is, null); |
| 696 |
|
|
| 693 |
|
|
| 697 | 694 |
config.makeNamespaces(model); |
| 698 |
|
|
| 695 |
|
|
| 699 | 696 |
String rdfNamespace = config.getRdfNamespaceURIString(); |
| 700 | 697 |
String taxonNameNamespace = config.getTnNamespaceURIString(); |
| 701 | 698 |
return handleNameModel(model, config, taxonNameMap, uri); |
| 702 |
|
|
| 703 |
|
|
| 699 |
|
|
| 700 |
|
|
| 704 | 701 |
}catch(Exception e){
|
| 705 | 702 |
logger.debug("The file was no valid rdf file");
|
| 706 | 703 |
} |
| 707 |
|
|
| 708 |
|
|
| 709 |
|
|
| 704 |
|
|
| 705 |
|
|
| 706 |
|
|
| 710 | 707 |
return null; |
| 711 | 708 |
} |
| 712 |
|
|
| 709 |
|
|
| 713 | 710 |
} |
Also available in: Unified diff
Revert "minor"
This reverts commit 64367d007ba404e037ccc2b6794b46c4f1f321bc.