Revision d2b6e0d7
Added by Katja Luther over 9 years ago
| cdm-pesi/src/main/java/eu/etaxonomy/cdm/io/pesi/indexFungorum/IndexFungorumSpeciesImport.java | ||
|---|---|---|
| 1 | 1 |
/** |
| 2 | 2 |
* Copyright (C) 2007 EDIT |
| 3 |
* European Distributed Institute of Taxonomy
|
|
| 3 |
* European Distributed Institute of Taxonomy |
|
| 4 | 4 |
* http://www.e-taxonomy.eu |
| 5 |
*
|
|
| 5 |
* |
|
| 6 | 6 |
* The contents of this file are subject to the Mozilla Public License Version 1.1 |
| 7 | 7 |
* See LICENSE.TXT at the top of this package for the full license terms. |
| 8 | 8 |
*/ |
| ... | ... | |
| 11 | 11 |
|
| 12 | 12 |
import java.sql.ResultSet; |
| 13 | 13 |
import java.sql.SQLException; |
| 14 |
import java.util.ArrayList; |
|
| 14 | 15 |
import java.util.HashMap; |
| 15 | 16 |
import java.util.HashSet; |
| 17 |
import java.util.List; |
|
| 16 | 18 |
import java.util.Map; |
| 17 | 19 |
import java.util.Set; |
| 20 |
import java.util.UUID; |
|
| 18 | 21 |
|
| 19 | 22 |
import org.apache.commons.lang.StringUtils; |
| 20 | 23 |
import org.apache.log4j.Logger; |
| 21 | 24 |
import org.springframework.stereotype.Component; |
| 22 | 25 |
|
| 26 |
import eu.etaxonomy.cdm.api.service.config.MatchingTaxonConfigurator; |
|
| 23 | 27 |
import eu.etaxonomy.cdm.common.CdmUtils; |
| 28 |
import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper; |
|
| 24 | 29 |
import eu.etaxonomy.cdm.io.common.ResultSetPartitioner; |
| 30 |
import eu.etaxonomy.cdm.io.common.Source; |
|
| 25 | 31 |
import eu.etaxonomy.cdm.io.pesi.erms.ErmsTransformer; |
| 26 | 32 |
import eu.etaxonomy.cdm.io.pesi.out.PesiTransformer; |
| 27 | 33 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
| 28 | 34 |
import eu.etaxonomy.cdm.model.common.Extension; |
| 29 | 35 |
import eu.etaxonomy.cdm.model.common.ExtensionType; |
| 30 |
import eu.etaxonomy.cdm.model.common.MarkerType; |
|
| 31 | 36 |
import eu.etaxonomy.cdm.model.name.NomenclaturalCode; |
| 32 | 37 |
import eu.etaxonomy.cdm.model.name.NonViralName; |
| 33 |
import eu.etaxonomy.cdm.model.name.Rank;
|
|
| 38 |
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
|
| 34 | 39 |
import eu.etaxonomy.cdm.model.reference.Reference; |
| 35 | 40 |
import eu.etaxonomy.cdm.model.taxon.Classification; |
| 36 | 41 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
| ... | ... | |
| 45 | 50 |
@Component |
| 46 | 51 |
public class IndexFungorumSpeciesImport extends IndexFungorumImportBase {
|
| 47 | 52 |
private static final Logger logger = Logger.getLogger(IndexFungorumSpeciesImport.class); |
| 48 |
|
|
| 53 |
|
|
| 49 | 54 |
private static final String pluralString = "species"; |
| 50 | 55 |
private static final String dbTableName = "[tblPESIfungi-IFdata]"; |
| 56 |
private final Set<UUID> infraspecificTaxaUUIDs = new HashSet<UUID>(); |
|
| 51 | 57 |
|
| 52 | 58 |
public IndexFungorumSpeciesImport(){
|
| 53 | 59 |
super(pluralString, dbTableName, null); |
| 60 |
|
|
| 54 | 61 |
} |
| 55 | 62 |
|
| 56 | 63 |
|
| 57 |
|
|
| 58 |
|
|
| 64 |
|
|
| 65 |
|
|
| 59 | 66 |
@Override |
| 60 | 67 |
protected String getIdQuery() {
|
| 61 | 68 |
String result = " SELECT PreferredNameIFnumber FROM " + getTableName() + |
| ... | ... | |
| 71 | 78 |
*/ |
| 72 | 79 |
@Override |
| 73 | 80 |
protected String getRecordQuery(IndexFungorumImportConfigurator config) {
|
| 74 |
String strRecordQuery =
|
|
| 81 |
String strRecordQuery = |
|
| 75 | 82 |
" SELECT DISTINCT distribution.PreferredNameFDCnumber, species.* , cl.[Phylum name]" + |
| 76 | 83 |
" FROM tblPESIfungi AS distribution RIGHT OUTER JOIN dbo.[tblPESIfungi-IFdata] AS species ON distribution.PreferredNameIFnumber = species.PreferredNameIFnumber " + |
| 77 | 84 |
" LEFT OUTER JOIN [tblPESIfungi-Classification] cl ON species.PreferredName = cl.PreferredName " + |
| 78 | 85 |
" WHERE ( species.PreferredNameIFnumber IN (" + ID_LIST_TOKEN + ") )" +
|
| 79 | 86 |
""; |
| 80 | 87 |
return strRecordQuery; |
| 88 |
} |
|
| 89 |
@Override |
|
| 90 |
protected void doInvoke(IndexFungorumImportState state){
|
|
| 91 |
System.out.println("start make " + getPluralString() + " ...");
|
|
| 92 |
IndexFungorumImportConfigurator config = state.getConfig(); |
|
| 93 |
Source source = config.getSource(); |
|
| 94 |
|
|
| 95 |
super.doInvoke(state); |
|
| 96 |
Classification classification = getClassification(state); |
|
| 97 |
List<TaxonBase> infraspecificTaxa = new ArrayList<TaxonBase>(); |
|
| 98 |
for (UUID uuid: infraspecificTaxaUUIDs){
|
|
| 99 |
infraspecificTaxa.add(getTaxonService().load(uuid)); |
|
| 100 |
} |
|
| 101 |
|
|
| 102 |
System.out.println("create infraspecific - specific relationship: " + infraspecificTaxa.size() + " taxa");
|
|
| 103 |
for (TaxonBase infraspecificTaxon: infraspecificTaxa){
|
|
| 104 |
HibernateProxyHelper.deproxy(infraspecificTaxon); |
|
| 105 |
TaxonNameBase name = infraspecificTaxon.getName(); |
|
| 106 |
|
|
| 107 |
UUID uuid = getNameService().saveOrUpdate(name); |
|
| 108 |
String parentNameString = getParentNameInfraSpecific(name); |
|
| 109 |
System.out.println("Parent name string: " + parentNameString);
|
|
| 110 |
MatchingTaxonConfigurator matchingConfig = new MatchingTaxonConfigurator(); |
|
| 111 |
matchingConfig.setTaxonNameTitle(parentNameString); |
|
| 112 |
List<String> propertyPaths = new ArrayList<String>(); |
|
| 113 |
propertyPaths.add("taxonNodes.*");
|
|
| 114 |
propertyPaths.add("taxonNodes.classification");
|
|
| 115 |
propertyPaths.add("taxonNodes.childNodes.*");
|
|
| 116 |
propertyPaths.add("taxonNodes.childNodes.taxon.*");
|
|
| 117 |
propertyPaths.add("taxonNodes.parent.*");
|
|
| 118 |
propertyPaths.add("taxonNodes.parent.taxon.*");
|
|
| 119 |
matchingConfig.setPropertyPath(propertyPaths); |
|
| 120 |
List<TaxonBase> potentialParents = getTaxonService().findTaxaByName(matchingConfig); |
|
| 121 |
//Taxon.class, parentNameString + "sec. ", MatchMode.BEGINNING, , pageSize, pageNumber, orderHints, propertyPaths) |
|
| 122 |
//.searchNames(String uninomial,String infraGenericEpithet, String specificEpithet, String infraspecificEpithet, Rank rank, Integer pageSize, Integer pageNumber, List<OrderHint> orderHints, |
|
| 123 |
if (potentialParents.size()>1){
|
|
| 124 |
for (TaxonBase potentialParent:potentialParents){
|
|
| 125 |
if (potentialParent.getTitleCache().equals(parentNameString + " sec*")){
|
|
| 126 |
classification.addParentChild((Taxon)potentialParent, (Taxon)infraspecificTaxon, null, null); |
|
| 127 |
} |
|
| 128 |
} |
|
| 129 |
}else if (!potentialParents.isEmpty()){
|
|
| 130 |
Taxon parent = HibernateProxyHelper.deproxy(potentialParents.get(0), Taxon.class); |
|
| 131 |
Taxon child = (Taxon)getTaxonService().load(infraspecificTaxon.getUuid(), propertyPaths); |
|
| 132 |
classification.addParentChild(parent, child, null, null); |
|
| 133 |
} else{
|
|
| 134 |
System.out.println("No parent for: " + name.getTitleCache());
|
|
| 135 |
} |
|
| 136 |
getTaxonService().saveOrUpdate(infraspecificTaxon); |
|
| 137 |
} |
|
| 138 |
|
|
| 139 |
|
|
| 81 | 140 |
} |
| 82 | 141 |
|
| 83 |
|
|
| 84 | 142 |
@Override |
| 85 | 143 |
public boolean doPartition(ResultSetPartitioner partitioner, IndexFungorumImportState state) {
|
| 86 | 144 |
boolean success = true; |
| 87 | 145 |
Reference<?> sourceReference = state.getRelatedObject(NAMESPACE_REFERENCE, SOURCE_REFERENCE, Reference.class); |
| 88 | 146 |
ResultSet rs = partitioner.getResultSet(); |
| 89 |
|
|
| 90 | 147 |
Classification classification = getClassification(state); |
| 148 |
|
|
| 91 | 149 |
try {
|
| 92 | 150 |
while (rs.next()){
|
| 93 | 151 |
|
| ... | ... | |
| 95 | 153 |
|
| 96 | 154 |
Integer id = (Integer)rs.getObject("PreferredNameIFnumber");
|
| 97 | 155 |
String phylumName = rs.getString("Phylum name");
|
| 98 |
|
|
| 156 |
|
|
| 99 | 157 |
String preferredName = rs.getString("PreferredName");
|
| 100 | 158 |
if (StringUtils.isBlank(preferredName)){
|
| 101 | 159 |
logger.warn("Preferred name is blank. This case is not yet handled by IF import. RECORD UMBER" + CdmUtils.Nz(id));
|
| 102 | 160 |
} |
| 103 |
|
|
| 104 |
Rank rank = Rank.SPECIES(); |
|
| 105 |
|
|
| 161 |
|
|
| 162 |
//Rank rank = Rank.SPECIES();
|
|
| 163 |
|
|
| 106 | 164 |
NonViralNameParserImpl parser = NonViralNameParserImpl.NewInstance(); |
| 107 |
NonViralName<?> name = parser.parseSimpleName(preferredName, NomenclaturalCode.ICNAFP, rank);
|
|
| 108 |
|
|
| 165 |
NonViralName<?> name = parser.parseSimpleName(preferredName, NomenclaturalCode.ICNAFP, null);
|
|
| 166 |
|
|
| 109 | 167 |
Taxon taxon = Taxon.NewInstance(name, sourceReference); |
| 110 |
Taxon parent = getParentTaxon(state, rs); |
|
| 111 |
if (parent == null){
|
|
| 112 |
logger.warn("parent not found for name:" +preferredName);
|
|
| 168 |
//if name is infraspecific the parent should be the species not the genus |
|
| 169 |
Taxon parent; |
|
| 170 |
if (!name.isInfraSpecific()){
|
|
| 171 |
parent = getParentTaxon(state, rs); |
|
| 172 |
if (parent == null){
|
|
| 173 |
logger.warn("parent not found for name:" +preferredName);
|
|
| 174 |
} |
|
| 175 |
classification.addParentChild(parent, taxon, null, null); |
|
| 113 | 176 |
} |
| 114 |
classification.addParentChild(parent, taxon, null, null); |
|
| 115 |
|
|
| 177 |
|
|
| 116 | 178 |
//author + publication |
| 117 | 179 |
makeAuthorAndPublication(state, rs, name); |
| 118 | 180 |
//source |
| 119 | 181 |
makeSource(state, taxon, id, NAMESPACE_SPECIES ); |
| 120 |
|
|
| 182 |
|
|
| 121 | 183 |
//fossil |
| 122 | 184 |
if (FOSSIL_FUNGI.equalsIgnoreCase(phylumName)){
|
| 123 | 185 |
ExtensionType fossilExtType = getExtensionType(state, ErmsTransformer.uuidFossilStatus, "fossil status", "fossil status", "fos. stat."); |
| 124 |
Extension.NewInstance(taxon, PesiTransformer.STR_FOSSIL_ONLY, fossilExtType);
|
|
| 186 |
Extension.NewInstance(taxon, PesiTransformer.STR_FOSSIL_ONLY, fossilExtType); |
|
| 125 | 187 |
} |
| 126 | 188 |
//save |
| 127 |
getTaxonService().saveOrUpdate(taxon); |
|
| 189 |
|
|
| 190 |
UUID uuidTaxon = getTaxonService().saveOrUpdate(taxon); |
|
| 191 |
getNameService().saveOrUpdate(name); |
|
| 192 |
if (name.isInfraSpecific()){
|
|
| 193 |
infraspecificTaxaUUIDs.add(uuidTaxon); |
|
| 194 |
} |
|
| 195 |
|
|
| 128 | 196 |
} |
| 129 | 197 |
|
| 130 |
|
|
| 198 |
|
|
| 131 | 199 |
} catch (Exception e) {
|
| 132 | 200 |
e.printStackTrace(); |
| 133 | 201 |
logger.error(e.getMessage()); |
| ... | ... | |
| 137 | 205 |
return success; |
| 138 | 206 |
} |
| 139 | 207 |
|
| 140 |
|
|
| 141 |
private Taxon getParentTaxon(IndexFungorumImportState state, ResultSet rs) throws SQLException {
|
|
| 208 |
|
|
| 209 |
/** |
|
| 210 |
* @param taxon |
|
| 211 |
* @return |
|
| 212 |
*/ |
|
| 213 |
private String getParentNameInfraSpecific(TaxonNameBase taxonName){
|
|
| 214 |
NonViralName<NonViralName> name = HibernateProxyHelper.deproxy(taxonName, NonViralName.class); |
|
| 215 |
String parentName = name.getGenusOrUninomial() + " " + name.getSpecificEpithet(); |
|
| 216 |
|
|
| 217 |
return parentName; |
|
| 218 |
} |
|
| 219 |
|
|
| 220 |
|
|
| 221 |
|
|
| 222 |
|
|
| 223 |
private Taxon getParentTaxon(IndexFungorumImportState state, ResultSet rs) throws SQLException {
|
|
| 142 | 224 |
Integer genusId = rs.getInt("PreferredNameFDCnumber");
|
| 143 |
|
|
| 225 |
|
|
| 144 | 226 |
Taxon taxon = state.getRelatedObject(NAMESPACE_GENERA, String.valueOf(genusId), Taxon.class); |
| 145 | 227 |
if (taxon == null){
|
| 146 | 228 |
logger.warn("Taxon not found for " + genusId);
|
| ... | ... | |
| 155 | 237 |
Class<?> cdmClass; |
| 156 | 238 |
Set<String> idSet; |
| 157 | 239 |
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<Object, Map<String, ? extends CdmBase>>(); |
| 158 |
|
|
| 240 |
|
|
| 159 | 241 |
try{
|
| 160 | 242 |
Set<String> taxonIdSet = new HashSet<String>(); |
| 161 |
while (rs.next()){
|
|
| 243 |
Set<String> taxonSpeciesNames = new HashSet<String>(); |
|
| 244 |
while (rs.next()){
|
|
| 162 | 245 |
handleForeignKey(rs, taxonIdSet,"PreferredNameFDCnumber" ); |
| 246 |
handleForeignKey(rs, taxonSpeciesNames, "PreferredName"); |
|
| 163 | 247 |
} |
| 164 |
|
|
| 248 |
|
|
| 165 | 249 |
//taxon map |
| 166 | 250 |
nameSpace = NAMESPACE_GENERA; |
| 167 | 251 |
cdmClass = TaxonBase.class; |
| 168 | 252 |
idSet = taxonIdSet; |
| 169 | 253 |
Map<String, TaxonBase> taxonMap = (Map<String, TaxonBase>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace); |
| 170 | 254 |
result.put(nameSpace, taxonMap); |
| 171 |
|
|
| 255 |
|
|
| 256 |
|
|
| 172 | 257 |
//sourceReference |
| 173 | 258 |
Reference<?> sourceReference = getReferenceService().find(PesiTransformer.uuidSourceRefIndexFungorum); |
| 174 | 259 |
Map<String, Reference> referenceMap = new HashMap<String, Reference>(); |
| 175 | 260 |
referenceMap.put(SOURCE_REFERENCE, sourceReference); |
| 176 | 261 |
result.put(NAMESPACE_REFERENCE, referenceMap); |
| 177 |
|
|
| 262 |
|
|
| 178 | 263 |
} catch (SQLException e) {
|
| 179 | 264 |
throw new RuntimeException(e); |
| 180 | 265 |
} |
| 181 | 266 |
return result; |
| 182 | 267 |
} |
| 183 |
|
|
| 268 |
|
|
| 184 | 269 |
@Override |
| 185 | 270 |
protected boolean doCheck(IndexFungorumImportState state){
|
| 186 | 271 |
return true; |
| 187 | 272 |
} |
| 188 |
|
|
| 273 |
|
|
| 189 | 274 |
@Override |
| 190 | 275 |
protected boolean isIgnore(IndexFungorumImportState state){
|
| 191 | 276 |
return ! state.getConfig().isDoTaxa(); |
Also available in: Unified diff
index fungorum import of infraspecific taxa