Revision c5e2b66a
Added by Andreas Müller 8 months ago
| app-import/src/main/java/eu/etaxonomy/cdm/app/caryophyllales/SileneaeImportActivator.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2023 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 |
package eu.etaxonomy.cdm.app.caryophyllales; |
|
| 10 |
|
|
| 11 |
import java.io.File; |
|
| 12 |
import java.util.UUID; |
|
| 13 |
|
|
| 14 |
import org.apache.logging.log4j.LogManager; |
|
| 15 |
import org.apache.logging.log4j.Logger; |
|
| 16 |
|
|
| 17 |
import eu.etaxonomy.cdm.app.common.CdmDestinations; |
|
| 18 |
import eu.etaxonomy.cdm.common.URI; |
|
| 19 |
import eu.etaxonomy.cdm.database.DbSchemaValidation; |
|
| 20 |
import eu.etaxonomy.cdm.database.ICdmDataSource; |
|
| 21 |
import eu.etaxonomy.cdm.io.caryo.CaryoSileneaeImportConfigurator; |
|
| 22 |
import eu.etaxonomy.cdm.io.common.CdmDefaultImport; |
|
| 23 |
import eu.etaxonomy.cdm.io.common.IImportConfigurator.CHECK; |
|
| 24 |
import eu.etaxonomy.cdm.model.reference.Reference; |
|
| 25 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
|
| 26 |
|
|
| 27 |
/** |
|
| 28 |
* @author a.mueller |
|
| 29 |
* @date 01.02.2023 |
|
| 30 |
*/ |
|
| 31 |
public class SileneaeImportActivator {
|
|
| 32 |
|
|
| 33 |
@SuppressWarnings("unused")
|
|
| 34 |
private static final Logger logger = LogManager.getLogger(); |
|
| 35 |
|
|
| 36 |
//database validation status (create, update, validate ...) |
|
| 37 |
private static final DbSchemaValidation hbm2dll = DbSchemaValidation.VALIDATE; |
|
| 38 |
private static final URI source = sileneae(); |
|
| 39 |
|
|
| 40 |
private static final ICdmDataSource cdmDestination = CdmDestinations.cdm_local_caryo_spp(); |
|
| 41 |
//private static final ICdmDataSource cdmDestination = CdmDestinations.cdm_production_caryophyllales_spp(); |
|
| 42 |
|
|
| 43 |
private static final UUID acceptedUuid = UUID.fromString("8ac4a182-1281-4473-b6a1-c4f681f7dea3");
|
|
| 44 |
private static final UUID unresolvedUuid = UUID.fromString("99ab0e67-4214-4bc7-884d-6af8c2e60c38");
|
|
| 45 |
private static final UUID classificationUuid = UUID.fromString("9edc58b5-de3b-43aa-9f31-1ede7c009c2b");
|
|
| 46 |
private static final UUID secUuid = UUID.fromString("770bf23b-68ab-4074-8241-3ed6899694af");
|
|
| 47 |
|
|
| 48 |
//check - import |
|
| 49 |
static CHECK check = CHECK.IMPORT_WITHOUT_CHECK; |
|
| 50 |
|
|
| 51 |
private void doImport(ICdmDataSource cdmDestination){
|
|
| 52 |
|
|
| 53 |
//make Source |
|
| 54 |
CaryoSileneaeImportConfigurator config = CaryoSileneaeImportConfigurator.NewInstance(source, cdmDestination); |
|
| 55 |
config.setClassificationUuid(classificationUuid); |
|
| 56 |
config.setAcceptedNodeUuid(acceptedUuid); |
|
| 57 |
config.setUnresolvedNodeUuid(unresolvedUuid); |
|
| 58 |
config.setCheck(check); |
|
| 59 |
// config.setDoTaxa(doTaxa); |
|
| 60 |
config.setDbSchemaValidation(hbm2dll); |
|
| 61 |
config.setSecUuid(secUuid); |
|
| 62 |
config.setSourceReferenceTitle("Sileneae.xlsx");
|
|
| 63 |
|
|
| 64 |
CdmDefaultImport<CaryoSileneaeImportConfigurator> myImport = new CdmDefaultImport<>(); |
|
| 65 |
|
|
| 66 |
//... |
|
| 67 |
if (true){
|
|
| 68 |
System.out.println("Start import from ("+ source.toString() + ") ...");
|
|
| 69 |
config.setSourceReference(getSourceReference(config.getSourceReferenceTitle())); |
|
| 70 |
myImport.invoke(config); |
|
| 71 |
System.out.println("End import from ("+ source.toString() + ")...");
|
|
| 72 |
} |
|
| 73 |
} |
|
| 74 |
|
|
| 75 |
private Reference getSourceReference(String string) {
|
|
| 76 |
Reference result = ReferenceFactory.newGeneric(); |
|
| 77 |
result.setTitleCache(string, true); |
|
| 78 |
return result; |
|
| 79 |
} |
|
| 80 |
|
|
| 81 |
public static URI sileneae(){
|
|
| 82 |
String fileName = "Sileneae.xlsx"; |
|
| 83 |
File file = new File("E://data/Caryophyllales/" + fileName);
|
|
| 84 |
if (!file.exists()) {
|
|
| 85 |
System.exit(0); |
|
| 86 |
} |
|
| 87 |
URI uri = URI.fromFile(file); |
|
| 88 |
return uri; |
|
| 89 |
} |
|
| 90 |
|
|
| 91 |
public static void main(String[] args) {
|
|
| 92 |
SileneaeImportActivator me = new SileneaeImportActivator(); |
|
| 93 |
me.doImport(cdmDestination); |
|
| 94 |
System.exit(0); |
|
| 95 |
} |
|
| 96 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeImportBase.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2023 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImport; |
|
| 12 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 13 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 14 |
import eu.etaxonomy.cdm.model.reference.Reference; |
|
| 15 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 16 |
|
|
| 17 |
/** |
|
| 18 |
* @author a.mueller |
|
| 19 |
* @date 02.02.2023 |
|
| 20 |
*/ |
|
| 21 |
public abstract class CaryoSileneaeImportBase extends SimpleExcelTaxonImport<CaryoSileneaeImportConfigurator> {
|
|
| 22 |
|
|
| 23 |
private static final long serialVersionUID = -299606747509423614L; |
|
| 24 |
|
|
| 25 |
private Reference secRef = null; |
|
| 26 |
|
|
| 27 |
protected Taxon getTaxon(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state, Integer taxonLinkId) {
|
|
| 28 |
// TODO Auto-generated method stub |
|
| 29 |
return null; |
|
| 30 |
} |
|
| 31 |
|
|
| 32 |
protected TaxonName getName(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state, Integer nameLinkID) {
|
|
| 33 |
// TODO Auto-generated method stub |
|
| 34 |
return null; |
|
| 35 |
} |
|
| 36 |
|
|
| 37 |
protected Reference getSecRef(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 38 |
if (secRef == null){
|
|
| 39 |
secRef = getReferenceService().find(state.getConfig().getSecUuid()); |
|
| 40 |
} |
|
| 41 |
return secRef; |
|
| 42 |
} |
|
| 43 |
|
|
| 44 |
protected void newTransaction(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 45 |
commitTransaction(state.getTransactionStatus()); |
|
| 46 |
secRef = null; |
|
| 47 |
state.getDeduplicationHelper().reset(); |
|
| 48 |
state.setSourceReference(null); |
|
| 49 |
System.gc(); |
|
| 50 |
state.setTransactionStatus(startTransaction()); |
|
| 51 |
} |
|
| 52 |
|
|
| 53 |
protected Integer getInt(String value) {
|
|
| 54 |
return isBlank(value) ? null : Integer.valueOf(value); |
|
| 55 |
} |
|
| 56 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeImportConfigurator.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2020 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import java.util.UUID; |
|
| 12 |
|
|
| 13 |
import eu.etaxonomy.cdm.common.URI; |
|
| 14 |
import eu.etaxonomy.cdm.database.ICdmDataSource; |
|
| 15 |
import eu.etaxonomy.cdm.io.common.ImportStateBase; |
|
| 16 |
import eu.etaxonomy.cdm.io.common.mapping.IInputTransformer; |
|
| 17 |
import eu.etaxonomy.cdm.io.excel.common.ExcelImportConfiguratorBase; |
|
| 18 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 19 |
import eu.etaxonomy.cdm.model.name.NomenclaturalCode; |
|
| 20 |
|
|
| 21 |
/** |
|
| 22 |
* @author a.mueller |
|
| 23 |
* @since 17.02.2020 |
|
| 24 |
*/ |
|
| 25 |
public class CaryoSileneaeImportConfigurator extends ExcelImportConfiguratorBase{
|
|
| 26 |
|
|
| 27 |
private static final long serialVersionUID = -3833210622605834032L; |
|
| 28 |
|
|
| 29 |
private boolean doDeduplicate = true; |
|
| 30 |
|
|
| 31 |
private UUID acceptedNodeUuid; |
|
| 32 |
private UUID unresolvedNodeUuid; |
|
| 33 |
|
|
| 34 |
private static IInputTransformer defaultTransformer = new CaryoAizoaceaeTransformer(); |
|
| 35 |
|
|
| 36 |
public static CaryoSileneaeImportConfigurator NewInstance(URI source, ICdmDataSource destination) {
|
|
| 37 |
return new CaryoSileneaeImportConfigurator(source, destination); |
|
| 38 |
} |
|
| 39 |
|
|
| 40 |
private CaryoSileneaeImportConfigurator(URI source, ICdmDataSource destination) {
|
|
| 41 |
super(source, destination, defaultTransformer); |
|
| 42 |
setNomenclaturalCode(NomenclaturalCode.ICNAFP); |
|
| 43 |
setSource(source); |
|
| 44 |
setDestination(destination); |
|
| 45 |
} |
|
| 46 |
|
|
| 47 |
@SuppressWarnings({ "unchecked", "rawtypes" })
|
|
| 48 |
@Override |
|
| 49 |
public ImportStateBase getNewState() {
|
|
| 50 |
return new SimpleExcelTaxonImportState<>(this); |
|
| 51 |
} |
|
| 52 |
|
|
| 53 |
@SuppressWarnings("unchecked")
|
|
| 54 |
@Override |
|
| 55 |
protected void makeIoClassList() {
|
|
| 56 |
ioClassList = new Class[]{
|
|
| 57 |
CaryoSileneaeNameImport.class |
|
| 58 |
}; |
|
| 59 |
} |
|
| 60 |
|
|
| 61 |
public boolean isDoDeduplicate() {
|
|
| 62 |
return this.doDeduplicate; |
|
| 63 |
} |
|
| 64 |
public void setDoDeduplicate(boolean doDeduplicate) {
|
|
| 65 |
this.doDeduplicate = doDeduplicate; |
|
| 66 |
} |
|
| 67 |
|
|
| 68 |
public UUID getAcceptedNodeUuid() {
|
|
| 69 |
return acceptedNodeUuid; |
|
| 70 |
} |
|
| 71 |
public void setAcceptedNodeUuid(UUID acceptedNodeUuid) {
|
|
| 72 |
this.acceptedNodeUuid = acceptedNodeUuid; |
|
| 73 |
} |
|
| 74 |
|
|
| 75 |
public UUID getUnresolvedNodeUuid() {
|
|
| 76 |
return unresolvedNodeUuid; |
|
| 77 |
} |
|
| 78 |
public void setUnresolvedNodeUuid(UUID unresolvedNodeUuid) {
|
|
| 79 |
this.unresolvedNodeUuid = unresolvedNodeUuid; |
|
| 80 |
} |
|
| 81 |
|
|
| 82 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeNameImport.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2020 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import java.util.HashMap; |
|
| 12 |
import java.util.List; |
|
| 13 |
import java.util.Map; |
|
| 14 |
import java.util.UUID; |
|
| 15 |
|
|
| 16 |
import org.apache.logging.log4j.LogManager; |
|
| 17 |
import org.apache.logging.log4j.Logger; |
|
| 18 |
import org.springframework.stereotype.Component; |
|
| 19 |
|
|
| 20 |
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade; |
|
| 21 |
import eu.etaxonomy.cdm.common.CdmUtils; |
|
| 22 |
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException; |
|
| 23 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 24 |
import eu.etaxonomy.cdm.model.common.Annotation; |
|
| 25 |
import eu.etaxonomy.cdm.model.common.AnnotationType; |
|
| 26 |
import eu.etaxonomy.cdm.model.name.NameTypeDesignationStatus; |
|
| 27 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType; |
|
| 28 |
import eu.etaxonomy.cdm.model.name.Rank; |
|
| 29 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus; |
|
| 30 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 31 |
import eu.etaxonomy.cdm.model.name.TaxonNameFactory; |
|
| 32 |
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationType; |
|
| 33 |
import eu.etaxonomy.cdm.model.reference.Reference; |
|
| 34 |
import eu.etaxonomy.cdm.model.term.DefinedTerm; |
|
| 35 |
import eu.etaxonomy.cdm.strategy.exceptions.StringNotParsableException; |
|
| 36 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl; |
|
| 37 |
|
|
| 38 |
/** |
|
| 39 |
* @author a.mueller |
|
| 40 |
* @since 02.02.2023 |
|
| 41 |
*/ |
|
| 42 |
@Component |
|
| 43 |
public class CaryoSileneaeNameImport extends CaryoSileneaeImportBase {
|
|
| 44 |
|
|
| 45 |
private static final long serialVersionUID = 8931253645038029899L; |
|
| 46 |
private static final Logger logger = LogManager.getLogger(); |
|
| 47 |
|
|
| 48 |
private static final String NOMEN_ID = "nomen_id"; |
|
| 49 |
private static final String RANK = "rank"; |
|
| 50 |
private static final String GENUS = "Genus"; |
|
| 51 |
private static final String INFRAGEN_NAME = "Infragen_name"; |
|
| 52 |
private static final String SPECIES = "Species"; |
|
| 53 |
private static final String INFRASP_EPITHET = "Infrasp_epithet"; |
|
| 54 |
private static final String AUTHORS = "Authors"; |
|
| 55 |
|
|
| 56 |
private static final String BASIONYM_LINK = "Basionym_link"; |
|
| 57 |
private static final String IPNI_ID = "IPNI_ID"; |
|
| 58 |
private static final String NOTES = "Notes"; |
|
| 59 |
private static final String NOMINVAL_FLAG = "NomInval_Flag"; |
|
| 60 |
private static final String TYPE_SPECIMEN = "Type_specimen"; |
|
| 61 |
private static final String TYPE_SPECIES_LINK = "TypeSpecies_link"; |
|
| 62 |
|
|
| 63 |
//maybe not used |
|
| 64 |
private static final String PUBLICATION = "Publication"; |
|
| 65 |
private static final String PUBL_DATE = "PublDate"; |
|
| 66 |
|
|
| 67 |
//not used |
|
| 68 |
@SuppressWarnings("unused")
|
|
| 69 |
private static final String HOMONYM_FLAG = "Homonym_Flag"; |
|
| 70 |
@SuppressWarnings("unused")
|
|
| 71 |
private static final String MISAPPLIED_FLAG = "Misapplied_Flag"; |
|
| 72 |
@SuppressWarnings("unused")
|
|
| 73 |
private static final String SynonymyReference_link = "SynonymyReference_link"; |
|
| 74 |
|
|
| 75 |
private Map<Integer, UUID> nameMapping = new HashMap<>(); |
|
| 76 |
|
|
| 77 |
private SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state; |
|
| 78 |
|
|
| 79 |
@Override |
|
| 80 |
protected void firstPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 81 |
int line = state.getCurrentLine(); |
|
| 82 |
if ((line % 500) == 0){
|
|
| 83 |
newTransaction(state); |
|
| 84 |
System.out.println(line); |
|
| 85 |
} |
|
| 86 |
|
|
| 87 |
this.state = state; |
|
| 88 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 89 |
|
|
| 90 |
Integer nomenId = Integer.valueOf(getValue(record, NOMEN_ID)); |
|
| 91 |
String rankStr = getValue(record, RANK); |
|
| 92 |
String genusStr = getValue(record, GENUS); |
|
| 93 |
String infragenStr = getValue(record, INFRAGEN_NAME); |
|
| 94 |
String speciesStr = getValue(record, SPECIES); |
|
| 95 |
String infraspStr = getValue(record, INFRASP_EPITHET); |
|
| 96 |
String authorsStr = getValue(record, AUTHORS); |
|
| 97 |
String ipniId = getValue(record, IPNI_ID); |
|
| 98 |
String notes = getValue(record, NOTES); |
|
| 99 |
String nomInvalFlag = getValue(record, NOMINVAL_FLAG); |
|
| 100 |
String typeSpecimenStr = getValue(record, TYPE_SPECIMEN); |
|
| 101 |
|
|
| 102 |
String row = String.valueOf(line) + "("+nomenId+"): ";
|
|
| 103 |
|
|
| 104 |
try {
|
|
| 105 |
//create name |
|
| 106 |
Rank rank = state.getTransformer().getRankByKey(rankStr); |
|
| 107 |
TaxonName name = TaxonNameFactory.NewBotanicalInstance(rank); |
|
| 108 |
|
|
| 109 |
//fill simple |
|
| 110 |
name.setGenusOrUninomial(Ne(genusStr)); |
|
| 111 |
name.setInfraGenericEpithet(Ne(infragenStr)); |
|
| 112 |
name.setSpecificEpithet(Ne(speciesStr)); |
|
| 113 |
name.setInfraSpecificEpithet(Ne(infraspStr)); |
|
| 114 |
NonViralNameParserImpl.NewInstance().parseAuthors(name, authorsStr); |
|
| 115 |
|
|
| 116 |
//TODO ??publication + PublDate |
|
| 117 |
|
|
| 118 |
//ipni ID |
|
| 119 |
if (isNotBlank(ipniId)) {
|
|
| 120 |
name.addIdentifier(ipniId, DefinedTerm.IDENTIFIER_NAME_IPNI()); |
|
| 121 |
} |
|
| 122 |
|
|
| 123 |
//notes |
|
| 124 |
if (isNotBlank(notes)) {
|
|
| 125 |
Annotation annotation = Annotation.NewDefaultLanguageInstance(notes); |
|
| 126 |
annotation.setAnnotationType(AnnotationType.TECHNICAL()); |
|
| 127 |
name.addAnnotation(annotation); |
|
| 128 |
} |
|
| 129 |
|
|
| 130 |
//nominval flag |
|
| 131 |
if (nomInvalFlag != null && nomInvalFlag.trim().equalsIgnoreCase("yes")) {
|
|
| 132 |
Reference ref = null; |
|
| 133 |
name.addStatus(NomenclaturalStatusType.INVALID(), ref, null); |
|
| 134 |
} |
|
| 135 |
|
|
| 136 |
//type specimen |
|
| 137 |
if (isNotBlank(typeSpecimenStr)) {
|
|
| 138 |
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(SpecimenOrObservationType.PreservedSpecimen); |
|
| 139 |
facade.setCollector(null); //just to create field unit and gathering event |
|
| 140 |
facade.innerDerivedUnit().setTitleCache(typeSpecimenStr, true); |
|
| 141 |
SpecimenTypeDesignationStatus status = null; //TODO |
|
| 142 |
Reference ref = null; |
|
| 143 |
String originalInfo = null; |
|
| 144 |
name.addSpecimenTypeDesignation(facade.innerDerivedUnit(), status, ref, null, originalInfo, false, false); |
|
| 145 |
// save ?? |
|
| 146 |
} |
|
| 147 |
|
|
| 148 |
|
|
| 149 |
} catch (UndefinedTransformerMethodException | StringNotParsableException e) {
|
|
| 150 |
e.printStackTrace(); |
|
| 151 |
} |
|
| 152 |
} |
|
| 153 |
|
|
| 154 |
private String Ne(String genusStr) {
|
|
| 155 |
return CdmUtils.Ne(genusStr); |
|
| 156 |
} |
|
| 157 |
|
|
| 158 |
private TaxonName dedupliateNameParts(TaxonName name) {
|
|
| 159 |
if (state.getConfig().isDoDeduplicate()){
|
|
| 160 |
state.getDeduplicationHelper().replaceAuthorNamesAndNomRef(name); |
|
| 161 |
} |
|
| 162 |
return name; |
|
| 163 |
} |
|
| 164 |
|
|
| 165 |
private String getOtherAuthors(List<TaxonName> otherNames) {
|
|
| 166 |
String result = ""; |
|
| 167 |
for (TaxonName name : otherNames){
|
|
| 168 |
result = CdmUtils.concat(";", result, name.getAuthorshipCache());
|
|
| 169 |
} |
|
| 170 |
return result; |
|
| 171 |
} |
|
| 172 |
|
|
| 173 |
|
|
| 174 |
private void handleNomenclRemarkAndNameStatus(String nomenclaturalRemarks, String row, boolean isNewName, TaxonName name, |
|
| 175 |
List<NomenclaturalStatusType> statusTypes) {
|
|
| 176 |
|
|
| 177 |
NomenclaturalStatusType remarkType = null; |
|
| 178 |
NomenclaturalStatusType statusType = statusTypes.isEmpty()? null: statusTypes.iterator().next(); |
|
| 179 |
if (nomenclaturalRemarks == null){
|
|
| 180 |
//nothing to do |
|
| 181 |
}else if (", nom. illeg.".equals(nomenclaturalRemarks)){
|
|
| 182 |
remarkType = NomenclaturalStatusType.ILLEGITIMATE(); |
|
| 183 |
}else if (", nom. cons.".equals(nomenclaturalRemarks)){
|
|
| 184 |
remarkType = NomenclaturalStatusType.CONSERVED(); |
|
| 185 |
}else if (", nom. nud.".equals(nomenclaturalRemarks)){
|
|
| 186 |
remarkType = NomenclaturalStatusType.NUDUM(); |
|
| 187 |
}else if (", nom. provis.".equals(nomenclaturalRemarks)){
|
|
| 188 |
remarkType = NomenclaturalStatusType.PROVISIONAL(); |
|
| 189 |
}else if (", nom. rej.".equals(nomenclaturalRemarks)){
|
|
| 190 |
remarkType = NomenclaturalStatusType.REJECTED(); |
|
| 191 |
}else if (", nom. subnud.".equals(nomenclaturalRemarks)){
|
|
| 192 |
remarkType = NomenclaturalStatusType.SUBNUDUM(); |
|
| 193 |
}else if (", nom. superfl.".equals(nomenclaturalRemarks)){
|
|
| 194 |
remarkType = NomenclaturalStatusType.SUPERFLUOUS(); |
|
| 195 |
}else if (", not validly publ.".equals(nomenclaturalRemarks)){
|
|
| 196 |
statusTypes.add(NomenclaturalStatusType.INVALID()); |
|
| 197 |
}else if (", opus utique oppr.".equals(nomenclaturalRemarks)){
|
|
| 198 |
statusTypes.add(NomenclaturalStatusType.OPUS_UTIQUE_OPPR()); |
|
| 199 |
}else {
|
|
| 200 |
logger.warn(row + "Unhandled nomenclatural remark: " + nomenclaturalRemarks); |
|
| 201 |
} |
|
| 202 |
|
|
| 203 |
NomenclaturalStatusType kewType = remarkType != null? remarkType : statusType; |
|
| 204 |
if (isNewName){
|
|
| 205 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
| 206 |
logger.warn(row + "Kew suggests 2 different nom. status. types for new name. The status from nomenclatural_remarks was taken."); |
|
| 207 |
} |
|
| 208 |
if (kewType != null){
|
|
| 209 |
name.addStatus(kewType, getSecRef(state), null); |
|
| 210 |
} |
|
| 211 |
}else{
|
|
| 212 |
NomenclaturalStatusType existingType = null; |
|
| 213 |
if (!name.getStatus().isEmpty()){
|
|
| 214 |
existingType = name.getStatus().iterator().next().getType(); |
|
| 215 |
} |
|
| 216 |
if (existingType != null && kewType != null){
|
|
| 217 |
if (!existingType.equals(kewType)){
|
|
| 218 |
logger.warn(row + "Existing name status "+existingType.getTitleCache()+" differs from Kew status " + kewType.getTitleCache() + ". Key status ignored"); |
|
| 219 |
} |
|
| 220 |
}else if (existingType != null && kewType == null){
|
|
| 221 |
logger.warn(row + "Info: Existing name has a name status "+existingType.getTitleCache()+" but Kew name has no status. Existing status kept."); |
|
| 222 |
}else if (existingType == null && kewType != null){
|
|
| 223 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
| 224 |
logger.warn(row + "Existing name has no status while Kew name suggests a status (but 2 different status form status and nomenclatural_remarks field)."); |
|
| 225 |
}else{
|
|
| 226 |
logger.warn(row + "Existing name has no status while Kew name suggests a status ("+kewType.getTitleCache()+"). Kew status ignored.");
|
|
| 227 |
} |
|
| 228 |
} |
|
| 229 |
} |
|
| 230 |
} |
|
| 231 |
|
|
| 232 |
@Override |
|
| 233 |
protected void secondPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 234 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 235 |
int line = state.getCurrentLine(); |
|
| 236 |
|
|
| 237 |
Integer nomenId = Integer.valueOf(getValue(record, NOMEN_ID)); |
|
| 238 |
Integer basionymId = getInt(getValue(record, BASIONYM_LINK)); |
|
| 239 |
Integer typeSpeciesId = getInt(getValue(record, TYPE_SPECIES_LINK)); |
|
| 240 |
|
|
| 241 |
String row = String.valueOf(line) + "("+nomenId+"): ";
|
|
| 242 |
if ((line % 500) == 0){
|
|
| 243 |
newTransaction(state); |
|
| 244 |
System.out.println(line); |
|
| 245 |
} |
|
| 246 |
|
|
| 247 |
TaxonName name = getName(state, nomenId); |
|
| 248 |
if (name == null) {
|
|
| 249 |
logger.warn(row + "Name does not exist"); |
|
| 250 |
return; |
|
| 251 |
} |
|
| 252 |
|
|
| 253 |
//basionym |
|
| 254 |
if (basionymId != null) {
|
|
| 255 |
TaxonName basionym = getName(state, basionymId); |
|
| 256 |
if (basionym == null) {
|
|
| 257 |
logger.warn(row + "basionym does not exist"); |
|
| 258 |
}else {
|
|
| 259 |
name.addBasionym(basionym); |
|
| 260 |
} |
|
| 261 |
} |
|
| 262 |
|
|
| 263 |
//type name |
|
| 264 |
if (typeSpeciesId != null) {
|
|
| 265 |
TaxonName typeSpecies = getName(state, typeSpeciesId); |
|
| 266 |
if (typeSpecies == null) {
|
|
| 267 |
logger.warn(row + "typeSpecies does not exist"); |
|
| 268 |
}else {
|
|
| 269 |
Reference ref = null; |
|
| 270 |
NameTypeDesignationStatus status = null; //TODO |
|
| 271 |
name.addNameTypeDesignation(typeSpecies, ref, null, null, status, false); |
|
| 272 |
} |
|
| 273 |
} |
|
| 274 |
} |
|
| 275 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeNomRefImport.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2020 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import java.util.HashMap; |
|
| 12 |
import java.util.HashSet; |
|
| 13 |
import java.util.List; |
|
| 14 |
import java.util.Map; |
|
| 15 |
import java.util.Set; |
|
| 16 |
import java.util.UUID; |
|
| 17 |
|
|
| 18 |
import org.apache.logging.log4j.LogManager; |
|
| 19 |
import org.apache.logging.log4j.Logger; |
|
| 20 |
import org.springframework.stereotype.Component; |
|
| 21 |
|
|
| 22 |
import eu.etaxonomy.cdm.common.CdmUtils; |
|
| 23 |
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException; |
|
| 24 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 25 |
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase; |
|
| 26 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 27 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType; |
|
| 28 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 29 |
import eu.etaxonomy.cdm.model.reference.Reference; |
|
| 30 |
import eu.etaxonomy.cdm.model.reference.ReferenceType; |
|
| 31 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
|
| 32 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 33 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl; |
|
| 34 |
|
|
| 35 |
/** |
|
| 36 |
* @author a.mueller |
|
| 37 |
* @since 02.02.2023 |
|
| 38 |
*/ |
|
| 39 |
@Component |
|
| 40 |
public class CaryoSileneaeNomRefImport extends CaryoSileneaeImportBase {
|
|
| 41 |
|
|
| 42 |
private static final long serialVersionUID = 7227226331297614469L; |
|
| 43 |
private static final Logger logger = LogManager.getLogger(); |
|
| 44 |
|
|
| 45 |
private static final String NOMEN_ID = "nomen_id"; |
|
| 46 |
private static final String NAME = "name"; |
|
| 47 |
private static final String PUBLICATION = "Publication"; |
|
| 48 |
private static final String PUB_TYPE_ED = "PubTypeEd"; |
|
| 49 |
private static final String PUB_TYPE_KEW = "PubTypeKew"; |
|
| 50 |
private static final String PUB_KEW = "PubKew"; |
|
| 51 |
private static final String NIMM_KEW = "NimmKew"; |
|
| 52 |
private static final String ORIG_SPELLING = "Original spelling"; |
|
| 53 |
private static final String NOM_STATUS = "Nom. Status"; |
|
| 54 |
|
|
| 55 |
private static final String SECOND_PUBLICATION = "SecondPublication"; |
|
| 56 |
private static final String IMPORT = "import"; |
|
| 57 |
private static final String DUPL = "dupl"; |
|
| 58 |
|
|
| 59 |
|
|
| 60 |
private Map<Integer, Reference> nameMapping = new HashMap<>(); |
|
| 61 |
private Set<String> neglectedRecords = new HashSet<>(); |
|
| 62 |
private Set<UUID> createdNames = new HashSet<>(); |
|
| 63 |
|
|
| 64 |
private static final NonViralNameParserImpl parser = NonViralNameParserImpl.NewInstance(); |
|
| 65 |
|
|
| 66 |
private SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state; |
|
| 67 |
|
|
| 68 |
@Override |
|
| 69 |
protected void firstPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 70 |
int line = state.getCurrentLine(); |
|
| 71 |
if ((line % 500) == 0){
|
|
| 72 |
newTransaction(state); |
|
| 73 |
System.out.println(line); |
|
| 74 |
} |
|
| 75 |
|
|
| 76 |
this.state = state; |
|
| 77 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 78 |
|
|
| 79 |
Integer nomenId = Integer.valueOf(getValue(record, NOMEN_ID)); |
|
| 80 |
String nameStr = getValue(record, NAME); |
|
| 81 |
String origPublication = getValue(record, PUBLICATION); |
|
| 82 |
String pubTypeEd = getValue(record, PUB_TYPE_ED); |
|
| 83 |
String pubTypeKew = getValue(record, PUB_TYPE_KEW); |
|
| 84 |
String pubKew = getValue(record, PUB_KEW); |
|
| 85 |
|
|
| 86 |
String nimmKew = getValue(record, NIMM_KEW); |
|
| 87 |
String origSpelling = getValue(record, ORIG_SPELLING); |
|
| 88 |
//TODO erstmal nicht importieren laut NaK |
|
| 89 |
@SuppressWarnings("unused")
|
|
| 90 |
String nomStatus = getValue(record, NOM_STATUS); |
|
| 91 |
|
|
| 92 |
String row = String.valueOf(line) + "("+nameStr+"): ";
|
|
| 93 |
|
|
| 94 |
TaxonName name = getName(state, nomenId); |
|
| 95 |
if (name == null) {
|
|
| 96 |
return; //record did not exist |
|
| 97 |
//TODO check if it is really a duplicate |
|
| 98 |
} |
|
| 99 |
|
|
| 100 |
boolean isKew = isNotBlank(nimmKew) && "x".equals(nimmKew); |
|
| 101 |
|
|
| 102 |
String publication = isKew ? pubKew : origPublication; |
|
| 103 |
String pubType = isKew ? pubTypeEd : pubTypeKew; |
|
| 104 |
|
|
| 105 |
ReferenceType refType = getRefType(pubType); |
|
| 106 |
if (refType == null) {
|
|
| 107 |
logger.warn(row + "reference type not found"); |
|
| 108 |
}else if (refType == ReferenceType.Article) {
|
|
| 109 |
publication = "in " + publication; |
|
| 110 |
}else if (refType == ReferenceType.Book) {
|
|
| 111 |
// |
|
| 112 |
}else {
|
|
| 113 |
logger.warn(row + "reference type not handled: " + refType); |
|
| 114 |
} |
|
| 115 |
String referenceName = name.getTitleCache()+ " " + publication; |
|
| 116 |
TaxonName parsedName = parser.parseReferencedName(referenceName); |
|
| 117 |
if (parsedName.isProtectedFullTitleCache() || parsedName.isProtectedTitleCache() ) {
|
|
| 118 |
logger.warn(row + "name could not be parsed"); |
|
| 119 |
}else {
|
|
| 120 |
Reference ref = parsedName.getNomenclaturalReference(); |
|
| 121 |
name.setNomenclaturalReference(ref); |
|
| 122 |
String microRef = parsedName.getNomenclaturalMicroReference(); |
|
| 123 |
name.setNomenclaturalMicroReference(microRef); |
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
//validateName (name); |
|
| 127 |
validateName(name, nameStr, row); |
|
| 128 |
//deduplicate |
|
| 129 |
dedupliateNameParts(name); |
|
| 130 |
|
|
| 131 |
//orig spelling |
|
| 132 |
if (isNotBlank(origSpelling)) {
|
|
| 133 |
TaxonName origName = (TaxonName)parser.parseFullName(origSpelling); |
|
| 134 |
if (origName.isProtectedTitleCache()) {
|
|
| 135 |
logger.warn(row + "orig name could not be parsed"); |
|
| 136 |
} |
|
| 137 |
name.getNomenclaturalSource().setNameUsedInSource(origName); |
|
| 138 |
} |
|
| 139 |
} |
|
| 140 |
|
|
| 141 |
private void validateName(TaxonName name, String nameStr, String row) {
|
|
| 142 |
if (name.getTitleCache().equals(nameStr)) {
|
|
| 143 |
logger.warn(row+ "name titleCache does not match"); |
|
| 144 |
} |
|
| 145 |
} |
|
| 146 |
|
|
| 147 |
private ReferenceType getRefType(String pubType) {
|
|
| 148 |
if ("A".equals(pubType)){
|
|
| 149 |
return ReferenceType.Article; |
|
| 150 |
}else if ("B".equals(pubType)) {
|
|
| 151 |
return ReferenceType.Book; |
|
| 152 |
} |
|
| 153 |
return null; |
|
| 154 |
} |
|
| 155 |
|
|
| 156 |
private TaxonName dedupliateNameParts(TaxonName name) {
|
|
| 157 |
if (state.getConfig().isDoDeduplicate()){
|
|
| 158 |
state.getDeduplicationHelper().replaceAuthorNamesAndNomRef(name); |
|
| 159 |
} |
|
| 160 |
return name; |
|
| 161 |
} |
|
| 162 |
|
|
| 163 |
private void makeStatus(String status, String sourceId, |
|
| 164 |
String accId, String row, List<NomenclaturalStatusType> statusTypes) {
|
|
| 165 |
|
|
| 166 |
Class<? extends CdmBase> clazz; |
|
| 167 |
if ("Accepted".equals(status) || "Unplaced".equals(status) || "Artificial Hybrid".equals(status) ){
|
|
| 168 |
clazz = Taxon.class; |
|
| 169 |
}else if ("Synonym".equals(status) || "Orthographic".equals(status)){
|
|
| 170 |
clazz = (accId == null)? Taxon.class : Synonym.class; |
|
| 171 |
if("Orthographic".equals(status)){
|
|
| 172 |
statusTypes.add(NomenclaturalStatusType.SUPERFLUOUS()); |
|
| 173 |
} |
|
| 174 |
}else if("Illegitimate".equals(status)){
|
|
| 175 |
statusTypes.add(NomenclaturalStatusType.ILLEGITIMATE()); |
|
| 176 |
}else if ("Invalid".equals(status)){
|
|
| 177 |
statusTypes.add(NomenclaturalStatusType.INVALID()); |
|
| 178 |
}else{
|
|
| 179 |
logger.warn(row + "Unhandled status: " + status); |
|
| 180 |
} |
|
| 181 |
return; |
|
| 182 |
} |
|
| 183 |
|
|
| 184 |
private void handleNomenclRemarkAndNameStatus(String nomenclaturalRemarks, String row, boolean isNewName, TaxonName name, |
|
| 185 |
List<NomenclaturalStatusType> statusTypes) {
|
|
| 186 |
|
|
| 187 |
NomenclaturalStatusType remarkType = null; |
|
| 188 |
NomenclaturalStatusType statusType = statusTypes.isEmpty()? null: statusTypes.iterator().next(); |
|
| 189 |
if (nomenclaturalRemarks == null){
|
|
| 190 |
//nothing to do |
|
| 191 |
}else if (", nom. illeg.".equals(nomenclaturalRemarks)){
|
|
| 192 |
remarkType = NomenclaturalStatusType.ILLEGITIMATE(); |
|
| 193 |
}else if (", nom. cons.".equals(nomenclaturalRemarks)){
|
|
| 194 |
remarkType = NomenclaturalStatusType.CONSERVED(); |
|
| 195 |
}else if (", nom. nud.".equals(nomenclaturalRemarks)){
|
|
| 196 |
remarkType = NomenclaturalStatusType.NUDUM(); |
|
| 197 |
}else if (", nom. provis.".equals(nomenclaturalRemarks)){
|
|
| 198 |
remarkType = NomenclaturalStatusType.PROVISIONAL(); |
|
| 199 |
}else if (", nom. rej.".equals(nomenclaturalRemarks)){
|
|
| 200 |
remarkType = NomenclaturalStatusType.REJECTED(); |
|
| 201 |
}else if (", nom. subnud.".equals(nomenclaturalRemarks)){
|
|
| 202 |
remarkType = NomenclaturalStatusType.SUBNUDUM(); |
|
| 203 |
}else if (", nom. superfl.".equals(nomenclaturalRemarks)){
|
|
| 204 |
remarkType = NomenclaturalStatusType.SUPERFLUOUS(); |
|
| 205 |
}else if (", not validly publ.".equals(nomenclaturalRemarks)){
|
|
| 206 |
statusTypes.add(NomenclaturalStatusType.INVALID()); |
|
| 207 |
}else if (", opus utique oppr.".equals(nomenclaturalRemarks)){
|
|
| 208 |
statusTypes.add(NomenclaturalStatusType.OPUS_UTIQUE_OPPR()); |
|
| 209 |
}else {
|
|
| 210 |
logger.warn(row + "Unhandled nomenclatural remark: " + nomenclaturalRemarks); |
|
| 211 |
} |
|
| 212 |
|
|
| 213 |
NomenclaturalStatusType kewType = remarkType != null? remarkType : statusType; |
|
| 214 |
if (isNewName){
|
|
| 215 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
| 216 |
logger.warn(row + "Kew suggests 2 different nom. status. types for new name. The status from nomenclatural_remarks was taken."); |
|
| 217 |
} |
|
| 218 |
if (kewType != null){
|
|
| 219 |
name.addStatus(kewType, getSecRef(state), null); |
|
| 220 |
} |
|
| 221 |
}else{
|
|
| 222 |
NomenclaturalStatusType existingType = null; |
|
| 223 |
if (!name.getStatus().isEmpty()){
|
|
| 224 |
existingType = name.getStatus().iterator().next().getType(); |
|
| 225 |
} |
|
| 226 |
if (existingType != null && kewType != null){
|
|
| 227 |
if (!existingType.equals(kewType)){
|
|
| 228 |
logger.warn(row + "Existing name status "+existingType.getTitleCache()+" differs from Kew status " + kewType.getTitleCache() + ". Key status ignored"); |
|
| 229 |
} |
|
| 230 |
}else if (existingType != null && kewType == null){
|
|
| 231 |
logger.warn(row + "Info: Existing name has a name status "+existingType.getTitleCache()+" but Kew name has no status. Existing status kept."); |
|
| 232 |
}else if (existingType == null && kewType != null){
|
|
| 233 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
| 234 |
logger.warn(row + "Existing name has no status while Kew name suggests a status (but 2 different status form status and nomenclatural_remarks field)."); |
|
| 235 |
}else{
|
|
| 236 |
logger.warn(row + "Existing name has no status while Kew name suggests a status ("+kewType.getTitleCache()+"). Kew status ignored.");
|
|
| 237 |
} |
|
| 238 |
} |
|
| 239 |
} |
|
| 240 |
} |
|
| 241 |
|
|
| 242 |
private void testParsedName(SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state, TaxonName name, |
|
| 243 |
String row, String fullCitation) throws UndefinedTransformerMethodException {
|
|
| 244 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 245 |
|
|
| 246 |
String nameCache = getValue(record, NAME); |
|
| 247 |
String pubType = getValue(record, "PubType"); |
|
| 248 |
String place_of_publication = getValue(record, "place_of_publication"); |
|
| 249 |
String volume_and_page = getValue(record, "volume_and_page"); |
|
| 250 |
|
|
| 251 |
// if (!CdmUtils.nullSafeEqual(name.getNameCache(), nameCache)){
|
|
| 252 |
// logger.warn(row + "Unexpected nameCache: " + nameCache); |
|
| 253 |
// } |
|
| 254 |
// if (!CdmUtils.nullSafeEqual(name.getTitleCache(), fullName)){
|
|
| 255 |
// logger.warn(row + "Unexpected titleCache: <->" + name.getTitleCache()); |
|
| 256 |
// } |
|
| 257 |
// if (isBlank(genusHybrid) == name.isMonomHybrid()){
|
|
| 258 |
// logger.warn(row + "Unexpected genus hybrid: " + genusHybrid); |
|
| 259 |
// } |
|
| 260 |
// if (!CdmUtils.nullSafeEqual(name.getGenusOrUninomial(),genus)){
|
|
| 261 |
// logger.warn(row + "Unexpected genus: " + genus); |
|
| 262 |
// }if (isBlank(speciesHybrid) == name.isBinomHybrid()){
|
|
| 263 |
// logger.warn(row + "Unexpected species hybrid: " + speciesHybrid); |
|
| 264 |
// } |
|
| 265 |
// if (!CdmUtils.nullSafeEqual(name.getSpecificEpithet(),species)){
|
|
| 266 |
// logger.warn(row + "Unexpected species epithet: " + name.getSpecificEpithet() +"<->"+ species); |
|
| 267 |
// } |
|
| 268 |
// if (!CdmUtils.nullSafeEqual(name.getInfraSpecificEpithet(), infraspecies)){
|
|
| 269 |
// logger.warn(row + "Unexpected infraspecific epithet: " + name.getInfraSpecificEpithet() +"<->"+ infraspecies); |
|
| 270 |
// } |
|
| 271 |
// if (!CdmUtils.nullSafeEqual(name.getAuthorshipCache(),authors)){
|
|
| 272 |
// logger.warn(row + "Unexpected authors: " + name.getAuthorshipCache() +"<->"+ authors); |
|
| 273 |
// } |
|
| 274 |
// String combinationAndExAuthor = authorTitle(name.getCombinationAuthorship(), name.getExCombinationAuthorship()); |
|
| 275 |
// if (!CdmUtils.nullSafeEqual(combinationAndExAuthor, combinationAuthor)){
|
|
| 276 |
// logger.warn(row + "Unexpected combination author: " + combinationAndExAuthor +"<->"+ combinationAuthor); |
|
| 277 |
// } |
|
| 278 |
// String basionymAndExAuthor = authorTitle(name.getBasionymAuthorship(), name.getExBasionymAuthorship()); |
|
| 279 |
// if (!CdmUtils.nullSafeEqual(basionymAndExAuthor, basionymAuthor)){
|
|
| 280 |
// logger.warn(row + "Unexpected basionym author: " + basionymAndExAuthor +"<->"+ basionymAuthor); |
|
| 281 |
// } |
|
| 282 |
// Rank rank = state.getTransformer().getRankByKey(rankStr); |
|
| 283 |
// if (!rank.equals(name.getRank())){
|
|
| 284 |
// logger.warn(row + "Unexpected rank: " + rankStr); |
|
| 285 |
// } |
|
| 286 |
// |
|
| 287 |
// Reference nomRef = name.getNomenclaturalReference(); |
|
| 288 |
// if (nomRef == null){
|
|
| 289 |
// if (fullCitation != null){
|
|
| 290 |
// NonViralNameParserImpl parser = new NonViralNameParserImpl(); |
|
| 291 |
// TaxonName parsedName = parser.parseReferencedName(fullCitation, NomenclaturalCode.ICNAFP, rank); |
|
| 292 |
// if (parsedName.getNomenclaturalReference() != null){
|
|
| 293 |
// name.setNomenclaturalReference(parsedName.getNomenclaturalReference()); |
|
| 294 |
// logger.warn(row + "Nom.ref. was missing. Taken from Kew"); |
|
| 295 |
// }else{
|
|
| 296 |
// logger.warn(row + "Nom. ref. is missing or can not be parsed"); |
|
| 297 |
// } |
|
| 298 |
// }else{
|
|
| 299 |
// logger.warn(row + "NomRef is missing."); |
|
| 300 |
// } |
|
| 301 |
// }else{
|
|
| 302 |
// if ("A".equals(pubType) && nomRef.getType() != ReferenceType.Article){
|
|
| 303 |
// logger.warn(row + "Unexpected nomref type: " + pubType + "<->" + nomRef.getType().toString()); |
|
| 304 |
// } |
|
| 305 |
// if ("B".equals(pubType) && nomRef.getType() != ReferenceType.Book){
|
|
| 306 |
// logger.warn(row + "Unexpected nomref type: " + pubType + "<->" + nomRef.getType().toString()); |
|
| 307 |
// } |
|
| 308 |
// year = normalizeYear(year); |
|
| 309 |
// if (!CdmUtils.nullSafeEqual(year, nomRef.getDatePublishedString())){
|
|
| 310 |
// logger.warn(row + "Unexpected year: " + year + "<->" + nomRef.getDatePublishedString()); |
|
| 311 |
// } |
|
| 312 |
// if (volume_and_page != null && !name.getFullTitleCache().contains(volume_and_page)){
|
|
| 313 |
// logger.warn(row + "volume_and_page not found in fullTitleCache: " + name.getFullTitleCache() +"<->"+ volume_and_page); |
|
| 314 |
// } |
|
| 315 |
// if (place_of_publication != null && !name.getFullTitleCache().contains(place_of_publication)){
|
|
| 316 |
// logger.warn(row + "place_of_publication not found in fullTitleCache: " + name.getFullTitleCache() +"<->"+ place_of_publication); |
|
| 317 |
// } |
|
| 318 |
// } |
|
| 319 |
// if ("subsp.".equals(infraSpecRank) && !rank.equals(Rank.SUBSPECIES())){
|
|
| 320 |
// logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank); |
|
| 321 |
// }else if ("var.".equals(infraSpecRank) && !rank.equals(Rank.VARIETY())){
|
|
| 322 |
// logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank); |
|
| 323 |
// }else if ("f.".equals(infraSpecRank) && !rank.equals(Rank.FORM())){
|
|
| 324 |
// logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank); |
|
| 325 |
// } |
|
| 326 |
} |
|
| 327 |
|
|
| 328 |
private String authorTitle(TeamOrPersonBase<?> author, TeamOrPersonBase<?> exAuthor) {
|
|
| 329 |
String authorStr = author == null? null: author.getNomenclaturalTitleCache(); |
|
| 330 |
String exAuthorStr = exAuthor == null? null: exAuthor.getNomenclaturalTitleCache(); |
|
| 331 |
return CdmUtils.concat(" ex ", exAuthorStr, authorStr);
|
|
| 332 |
} |
|
| 333 |
|
|
| 334 |
private String normalizeYear(String year) {
|
|
| 335 |
if (year == null){
|
|
| 336 |
return null; |
|
| 337 |
}else if (year.contains("\" [")){
|
|
| 338 |
String[] split = year.split("\" \\[");
|
|
| 339 |
year = split[1].replace("]","") + " [" + split[0]+"\"]";
|
|
| 340 |
}else if ("?".equals(year)){
|
|
| 341 |
return null; |
|
| 342 |
} |
|
| 343 |
return year; |
|
| 344 |
} |
|
| 345 |
|
|
| 346 |
@Override |
|
| 347 |
protected void secondPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 348 |
// Map<String, String> record = state.getOriginalRecord(); |
|
| 349 |
// int line = state.getCurrentLine(); |
|
| 350 |
// String accName = getValue(record, "AcceptedName"); |
|
| 351 |
// String basionymId = getValue(record, "basionym_plant_name_id"); |
|
| 352 |
// String homotypicSynonym = getValue(record, "homotypic_synonym"); |
|
| 353 |
// |
|
| 354 |
// String row = String.valueOf(line) + "("+fullName+"): ";
|
|
| 355 |
// |
|
| 356 |
// if ((line % 500) == 0){
|
|
| 357 |
// newTransaction(state); |
|
| 358 |
// System.out.println(line); |
|
| 359 |
// } |
|
| 360 |
} |
|
| 361 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeSynonymImport.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2020 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import java.util.HashSet; |
|
| 12 |
import java.util.Map; |
|
| 13 |
import java.util.Set; |
|
| 14 |
|
|
| 15 |
import org.apache.logging.log4j.LogManager; |
|
| 16 |
import org.apache.logging.log4j.Logger; |
|
| 17 |
import org.springframework.stereotype.Component; |
|
| 18 |
|
|
| 19 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 20 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 21 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 22 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
|
| 23 |
import eu.etaxonomy.cdm.model.taxon.SynonymType; |
|
| 24 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 25 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
| 26 |
|
|
| 27 |
/** |
|
| 28 |
* @author a.mueller |
|
| 29 |
* @since 02.02.2023 |
|
| 30 |
*/ |
|
| 31 |
@Component |
|
| 32 |
public class CaryoSileneaeSynonymImport extends CaryoSileneaeImportBase {
|
|
| 33 |
|
|
| 34 |
private static final long serialVersionUID = 7967768097472488888L; |
|
| 35 |
private static final Logger logger = LogManager.getLogger(); |
|
| 36 |
|
|
| 37 |
private static final String NOMTAX_ID = "NomTax_ID"; |
|
| 38 |
private static final String NOMEN_LINK = "Nomen_link"; |
|
| 39 |
private static final String TAXON_LINK = "Taxon_link"; |
|
| 40 |
|
|
| 41 |
private Set<String> neglectedRecords = new HashSet<>(); |
|
| 42 |
|
|
| 43 |
private SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state; |
|
| 44 |
|
|
| 45 |
@Override |
|
| 46 |
protected void firstPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 47 |
int line = state.getCurrentLine(); |
|
| 48 |
if ((line % 500) == 0){
|
|
| 49 |
newTransaction(state); |
|
| 50 |
System.out.println(line); |
|
| 51 |
} |
|
| 52 |
|
|
| 53 |
this.state = state; |
|
| 54 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 55 |
|
|
| 56 |
Integer nomTaxId = Integer.valueOf(getValue(record, NOMTAX_ID)); |
|
| 57 |
Integer nameLinkID = Integer.valueOf(getValue(record, NOMEN_LINK)); |
|
| 58 |
Integer taxonLinkId = Integer.valueOf(getValue(record, TAXON_LINK)); |
|
| 59 |
|
|
| 60 |
|
|
| 61 |
String row = String.valueOf(line) + "("+nomTaxId+"): ";
|
|
| 62 |
|
|
| 63 |
TaxonName name = getName(state, nameLinkID); |
|
| 64 |
|
|
| 65 |
if (name == null) {
|
|
| 66 |
logger.warn(row + "Name does not exist"); |
|
| 67 |
return; |
|
| 68 |
} |
|
| 69 |
|
|
| 70 |
Synonym synonym = Synonym.NewInstance(name, getSecRef(state)); |
|
| 71 |
|
|
| 72 |
Taxon taxon = getTaxon(state, taxonLinkId); |
|
| 73 |
if (taxon == null) {
|
|
| 74 |
logger.warn(row + "Taxon does not exist"); |
|
| 75 |
return; |
|
| 76 |
} |
|
| 77 |
//TODO type (compute homotypics) |
|
| 78 |
taxon.addSynonym(synonym, SynonymType.SYNONYM_OF); |
|
| 79 |
} |
|
| 80 |
|
|
| 81 |
@Override |
|
| 82 |
protected void secondPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 83 |
} |
|
| 84 |
|
|
| 85 |
private boolean hasSameAcceptedTaxon(TaxonBase<?> taxonBase, TaxonBase<?> basionymTaxon) {
|
|
| 86 |
if (taxonBase.isInstanceOf(Synonym.class)){
|
|
| 87 |
taxonBase = CdmBase.deproxy(taxonBase, Synonym.class).getAcceptedTaxon(); |
|
| 88 |
} |
|
| 89 |
if (basionymTaxon.isInstanceOf(Synonym.class)){
|
|
| 90 |
basionymTaxon = CdmBase.deproxy(basionymTaxon, Synonym.class).getAcceptedTaxon(); |
|
| 91 |
} |
|
| 92 |
return taxonBase != null && basionymTaxon != null && taxonBase.equals(basionymTaxon); |
|
| 93 |
} |
|
| 94 |
} |
|
| app-import/src/main/java/eu/etaxonomy/cdm/io/caryo/CaryoSileneaeTaxonImport.java | ||
|---|---|---|
| 1 |
/** |
|
| 2 |
* Copyright (C) 2020 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 |
package eu.etaxonomy.cdm.io.caryo; |
|
| 10 |
|
|
| 11 |
import java.util.HashMap; |
|
| 12 |
import java.util.HashSet; |
|
| 13 |
import java.util.Map; |
|
| 14 |
import java.util.Set; |
|
| 15 |
import java.util.UUID; |
|
| 16 |
|
|
| 17 |
import org.apache.logging.log4j.LogManager; |
|
| 18 |
import org.apache.logging.log4j.Logger; |
|
| 19 |
import org.springframework.stereotype.Component; |
|
| 20 |
|
|
| 21 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState; |
|
| 22 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 23 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 24 |
import eu.etaxonomy.cdm.model.taxon.Classification; |
|
| 25 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
|
| 26 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 27 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
| 28 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode; |
|
| 29 |
|
|
| 30 |
/** |
|
| 31 |
* @author a.mueller |
|
| 32 |
* @since 02.02.2023 |
|
| 33 |
*/ |
|
| 34 |
@Component |
|
| 35 |
public class CaryoSileneaeTaxonImport extends CaryoSileneaeImportBase {
|
|
| 36 |
|
|
| 37 |
private static final long serialVersionUID = 5594951908819469636L; |
|
| 38 |
private static final Logger logger = LogManager.getLogger(); |
|
| 39 |
|
|
| 40 |
private static final String ACCEPTED_TAXON_ID = "AcceptedTaxon_ID"; |
|
| 41 |
private static final String NOMEN_LINK = "nomen_link"; |
|
| 42 |
private static final String PARENT = "parent"; |
|
| 43 |
|
|
| 44 |
private Map<String, UUID> taxonMapping = new HashMap<>(); |
|
| 45 |
private Set<String> neglectedRecords = new HashSet<>(); |
|
| 46 |
|
|
| 47 |
private SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state; |
|
| 48 |
|
|
| 49 |
@Override |
|
| 50 |
protected void firstPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 51 |
int line = state.getCurrentLine(); |
|
| 52 |
if ((line % 500) == 0){
|
|
| 53 |
newTransaction(state); |
|
| 54 |
System.out.println(line); |
|
| 55 |
} |
|
| 56 |
|
|
| 57 |
this.state = state; |
|
| 58 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 59 |
|
|
| 60 |
Integer accTaxonId = Integer.valueOf(getValue(record, ACCEPTED_TAXON_ID)); |
|
| 61 |
Integer nameLinkID = Integer.valueOf(getValue(record, NOMEN_LINK)); |
|
| 62 |
|
|
| 63 |
String row = String.valueOf(line) + "("+accTaxonId+"): ";
|
|
| 64 |
|
|
| 65 |
TaxonName name = getName(state, nameLinkID); |
|
| 66 |
|
|
| 67 |
if (name == null) {
|
|
| 68 |
logger.warn(row + "Name does not exist"); |
|
| 69 |
return; |
|
| 70 |
} |
|
| 71 |
|
|
| 72 |
Taxon taxon = Taxon.NewInstance(name, getSecRef(state)); |
|
| 73 |
|
|
| 74 |
getTaxonService().saveOrUpdate(taxon); |
|
| 75 |
} |
|
| 76 |
|
|
| 77 |
@Override |
|
| 78 |
protected void secondPass(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state) {
|
|
| 79 |
int line = state.getCurrentLine(); |
|
| 80 |
if ((line % 500) == 0){
|
|
| 81 |
newTransaction(state); |
|
| 82 |
System.out.println(line); |
|
| 83 |
} |
|
| 84 |
|
|
| 85 |
this.state = state; |
|
| 86 |
Map<String, String> record = state.getOriginalRecord(); |
|
| 87 |
|
|
| 88 |
Integer accTaxonId = Integer.valueOf(getValue(record, ACCEPTED_TAXON_ID)); |
|
| 89 |
Integer parentId = Integer.valueOf(getValue(record, PARENT)); |
|
| 90 |
|
|
| 91 |
String row = String.valueOf(line) + "("+accTaxonId+"): ";
|
|
| 92 |
|
|
| 93 |
Taxon taxon = getTaxon(state, accTaxonId); |
|
| 94 |
if (taxon == null) {
|
|
| 95 |
logger.warn(row + "Taxon does not exist"); |
|
| 96 |
return; |
|
| 97 |
} |
|
| 98 |
|
|
| 99 |
Taxon parent = getTaxon(state, parentId); |
|
| 100 |
if (parent == null) {
|
|
| 101 |
logger.warn(row + "Taxon does not exist"); |
|
| 102 |
return; |
|
| 103 |
} |
|
| 104 |
|
|
| 105 |
Classification classification = getClassification(state); |
|
| 106 |
|
|
| 107 |
//TODO ref needed? |
|
| 108 |
TaxonNode node = classification.addParentChild(parent, taxon, null, null); |
|
| 109 |
|
|
| 110 |
getTaxonNodeService().saveOrUpdate(node); |
|
| 111 |
} |
|
| 112 |
|
|
| 113 |
private Classification getClassification(SimpleExcelTaxonImportState<CaryoSileneaeImportConfigurator> state2) {
|
|
| 114 |
// TODO Auto-generated method stub |
|
| 115 |
return null; |
|
| 116 |
} |
|
| 117 |
|
|
| 118 |
private boolean hasSameAcceptedTaxon(TaxonBase<?> taxonBase, TaxonBase<?> basionymTaxon) {
|
|
| 119 |
if (taxonBase.isInstanceOf(Synonym.class)){
|
|
| 120 |
taxonBase = CdmBase.deproxy(taxonBase, Synonym.class).getAcceptedTaxon(); |
|
| 121 |
} |
|
| 122 |
if (basionymTaxon.isInstanceOf(Synonym.class)){
|
|
| 123 |
basionymTaxon = CdmBase.deproxy(basionymTaxon, Synonym.class).getAcceptedTaxon(); |
|
| 124 |
} |
|
| 125 |
return taxonBase != null && basionymTaxon != null && taxonBase.equals(basionymTaxon); |
|
| 126 |
} |
|
| 127 |
|
|
| 128 |
private TaxonNode getFamily(){
|
|
| 129 |
UUID uuid = UUID.fromString("0334809a-aa20-447d-add9-138194f80f56");
|
|
| 130 |
TaxonNode aizoaceae = getTaxonNodeService().find(uuid); |
|
| 131 |
return aizoaceae; |
|
| 132 |
} |
|
| 133 |
|
|
| 134 |
private TaxonNode hybridParent(){
|
|
| 135 |
UUID uuid = UUID.fromString("2fae0fa1-758a-4fcb-bb6c-a2bd11f40641");
|
|
| 136 |
TaxonNode hybridParent = getTaxonNodeService().find(uuid); |
|
| 137 |
return hybridParent; |
|
| 138 |
} |
|
| 139 |
private TaxonNode unresolvedParent(){
|
|
| 140 |
UUID uuid = UUID.fromString("accb1ff6-5748-4b18-b529-9368c331a38d");
|
|
| 141 |
TaxonNode unresolvedParent = getTaxonNodeService().find(uuid); |
|
| 142 |
return unresolvedParent; |
|
| 143 |
} |
|
| 144 |
} |
|
Also available in: Unified diff
ref #10242 first version of Sileneae import