Revision b5b0fcf5
Added by Andreas Müller over 4 years ago
| cdm-pesi/src/main/java/eu/etaxonomy/cdm/app/pesi/merging/FaunaEuErmsFindIdenticalNamesActivator.java | ||
|---|---|---|
| 1 |
package eu.etaxonomy.cdm.app.pesi.merging; |
|
| 2 |
|
|
| 3 |
import java.io.FileWriter; |
|
| 4 |
import java.io.IOException; |
|
| 5 |
import java.util.ArrayList; |
|
| 6 |
import java.util.Iterator; |
|
| 7 |
import java.util.List; |
|
| 8 |
import java.util.Map; |
|
| 9 |
import java.util.Set; |
|
| 10 |
import java.util.UUID; |
|
| 11 |
|
|
| 12 |
import org.apache.log4j.Logger; |
|
| 13 |
|
|
| 14 |
import eu.etaxonomy.cdm.api.application.CdmApplicationController; |
|
| 15 |
import eu.etaxonomy.cdm.app.common.CdmDestinations; |
|
| 16 |
import eu.etaxonomy.cdm.database.DbSchemaValidation; |
|
| 17 |
import eu.etaxonomy.cdm.database.ICdmDataSource; |
|
| 18 |
import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper; |
|
| 19 |
import eu.etaxonomy.cdm.io.api.application.CdmIoApplicationController; |
|
| 20 |
import eu.etaxonomy.cdm.io.pesi.merging.PesiMergeObject; |
|
| 21 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 22 |
import eu.etaxonomy.cdm.model.common.IdentifiableSource; |
|
| 23 |
import eu.etaxonomy.cdm.model.name.IZoologicalName; |
|
| 24 |
import eu.etaxonomy.cdm.model.name.Rank; |
|
| 25 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 26 |
import eu.etaxonomy.cdm.model.reference.Reference; |
|
| 27 |
import eu.etaxonomy.cdm.model.taxon.Classification; |
|
| 28 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
|
| 29 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 30 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
| 31 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode; |
|
| 32 |
import eu.etaxonomy.cdm.persistence.dto.TaxonNodeDto; |
|
| 33 |
|
|
| 34 |
public class FaunaEuErmsFindIdenticalNamesActivator {
|
|
| 35 |
|
|
| 36 |
private static final Logger logger = Logger.getLogger(FaunaEuErmsFindIdenticalNamesActivator.class); |
|
| 37 |
|
|
| 38 |
//static final ICdmDataSource faunaEuropaeaSource = CdmDestinations.localH2(); |
|
| 39 |
static final ICdmDataSource pesiSource = CdmDestinations.cdm_test_local_faunaEu_mysql(); |
|
| 40 |
|
|
| 41 |
static final String path = "C:\\Users\\k.luther\\test"; |
|
| 42 |
|
|
| 43 |
private static UUID faunaEuSecUuid = UUID.fromString("6786d863-75d4-4796-b916-c1c3dff4cb70");
|
|
| 44 |
private static UUID ermsSecUuid = UUID.fromString("7744bc26-f914-42c4-b54a-dd2a030a8bb7");
|
|
| 45 |
private static UUID ifSecUuid; |
|
| 46 |
private static UUID emSecUuid; |
|
| 47 |
|
|
| 48 |
private void invoke(ICdmDataSource source){
|
|
| 49 |
|
|
| 50 |
CdmApplicationController app = CdmIoApplicationController.NewInstance(source, DbSchemaValidation.VALIDATE, false); |
|
| 51 |
|
|
| 52 |
List<String> propertyPaths = new ArrayList<>(); |
|
| 53 |
propertyPaths.add("sources.*");
|
|
| 54 |
propertyPaths.add("sources.idInSource");
|
|
| 55 |
propertyPaths.add("sources.idNamespace");
|
|
| 56 |
propertyPaths.add("taxonBases.*");
|
|
| 57 |
propertyPaths.add("taxonBases.relationsFromThisTaxon");
|
|
| 58 |
propertyPaths.add("taxonBases.taxonNodes.*");
|
|
| 59 |
propertyPaths.add("taxonBases.taxonNodes.parent.*");
|
|
| 60 |
propertyPaths.add("taxonBases.taxonNodes.childNodes.*");
|
|
| 61 |
propertyPaths.add("taxonBases.taxonNodes.childNodes.classification.rootNode.childNodes.*");
|
|
| 62 |
propertyPaths.add("taxonBases.taxonNodes.parent.taxon.name.*");
|
|
| 63 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.*");
|
|
| 64 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.childNodes.*");
|
|
| 65 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.childNodes.classification.rootNode.childNodes.*");
|
|
| 66 |
System.out.println("Start getIdenticalNames...");
|
|
| 67 |
|
|
| 68 |
Reference faunaEuSec = app.getReferenceService().load(faunaEuSecUuid); |
|
| 69 |
Reference ermsSec = app.getReferenceService().load(ermsSecUuid); |
|
| 70 |
Map<String, List<TaxonName>> namesOfIdenticalTaxa = app.getTaxonService().findIdenticalTaxonNameIds(ermsSec, faunaEuSec, propertyPaths); |
|
| 71 |
|
|
| 72 |
System.out.println("Start creating merging objects");
|
|
| 73 |
List<PesiMergeObject> mergingObjects= createMergeObjects(namesOfIdenticalTaxa, app); |
|
| 74 |
boolean resultOK = true; |
|
| 75 |
System.out.println("Start creating csv files");
|
|
| 76 |
resultOK &= writeSameNamesDifferentAuthorToCsv(mergingObjects, path + "_authors.csv"); |
|
| 77 |
resultOK &= writeSameNamesDifferentStatusToCsv(mergingObjects, path + "_status.csv"); |
|
| 78 |
resultOK &= writeSameNamesToCsvFile(mergingObjects, path + "_names.csv"); |
|
| 79 |
resultOK &= writeSameNamesDifferentPhylumToCsv(mergingObjects, path + "_phylum.csv"); |
|
| 80 |
resultOK &= writeSameNamesDifferentParentToCsv(mergingObjects, path + "parent.csv"); |
|
| 81 |
|
|
| 82 |
System.out.println("End merging Fauna Europaea and Erms: " + resultOK);
|
|
| 83 |
} |
|
| 84 |
|
|
| 85 |
private boolean writeSameNamesToCsvFile( |
|
| 86 |
List<PesiMergeObject> mergingObjects, String string) {
|
|
| 87 |
|
|
| 88 |
try{
|
|
| 89 |
FileWriter writer = new FileWriter(string); |
|
| 90 |
|
|
| 91 |
//create Header |
|
| 92 |
String firstLine = "same names"; |
|
| 93 |
createHeader(writer, firstLine); |
|
| 94 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 95 |
writeCsvLine(writer, merging) ; |
|
| 96 |
} |
|
| 97 |
writer.flush(); |
|
| 98 |
writer.close(); |
|
| 99 |
return true; |
|
| 100 |
}catch(IOException e){
|
|
| 101 |
logger.error(e.getMessage()); |
|
| 102 |
return false; |
|
| 103 |
} |
|
| 104 |
} |
|
| 105 |
|
|
| 106 |
private boolean writeSameNamesDifferentPhylumToCsv( |
|
| 107 |
List<PesiMergeObject> mergingObjects, String sfileName){
|
|
| 108 |
|
|
| 109 |
try{
|
|
| 110 |
FileWriter writer = new FileWriter(sfileName); |
|
| 111 |
|
|
| 112 |
//create Header |
|
| 113 |
String firstLine = "same names but different phylum"; |
|
| 114 |
createHeader(writer, firstLine); |
|
| 115 |
|
|
| 116 |
//write data |
|
| 117 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 118 |
//TODO |
|
| 119 |
if ((merging.getPhylumInErms()== null )^ (merging.getPhylumInFaunaEu()== null)){
|
|
| 120 |
writeCsvLine(writer, merging) ; |
|
| 121 |
}else if(!((merging.getPhylumInErms()==null) && (merging.getPhylumInFaunaEu()==null))){
|
|
| 122 |
if(!merging.getPhylumInErms().equals(merging.getPhylumInFaunaEu())){
|
|
| 123 |
writeCsvLine(writer, merging) ; |
|
| 124 |
} |
|
| 125 |
} |
|
| 126 |
} |
|
| 127 |
writer.flush(); |
|
| 128 |
writer.close(); |
|
| 129 |
return true; |
|
| 130 |
}catch(IOException e){
|
|
| 131 |
logger.error(e.getMessage()); |
|
| 132 |
return false; |
|
| 133 |
} |
|
| 134 |
} |
|
| 135 |
|
|
| 136 |
private boolean writeSameNamesDifferentParentToCsv( |
|
| 137 |
List<PesiMergeObject> mergingObjects, String sfileName){
|
|
| 138 |
|
|
| 139 |
try{
|
|
| 140 |
FileWriter writer = new FileWriter(sfileName); |
|
| 141 |
|
|
| 142 |
//create Header |
|
| 143 |
String firstLine = "same names but different parent"; |
|
| 144 |
createHeader(writer, firstLine); |
|
| 145 |
|
|
| 146 |
//write data |
|
| 147 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 148 |
//TODO |
|
| 149 |
if ((merging.getParentStringInErms()== null )^ (merging.getParentStringInFaunaEu()== null)){
|
|
| 150 |
writeCsvLine(writer, merging) ; |
|
| 151 |
}else if(!((merging.getParentStringInErms()==null) && (merging.getParentStringInFaunaEu()==null))){
|
|
| 152 |
if(!merging.getParentStringInErms().equals(merging.getParentStringInFaunaEu())){
|
|
| 153 |
writeCsvLine(writer, merging) ; |
|
| 154 |
} |
|
| 155 |
} |
|
| 156 |
} |
|
| 157 |
writer.flush(); |
|
| 158 |
writer.close(); |
|
| 159 |
return true; |
|
| 160 |
}catch(IOException e){
|
|
| 161 |
return false; |
|
| 162 |
} |
|
| 163 |
} |
|
| 164 |
|
|
| 165 |
private boolean writeSameNamesdifferentRankToCsv( |
|
| 166 |
List<PesiMergeObject> mergingObjects, String sfileName){
|
|
| 167 |
|
|
| 168 |
try{
|
|
| 169 |
FileWriter writer = new FileWriter(sfileName); |
|
| 170 |
String firstLine = "same names but different rank"; |
|
| 171 |
//create Header |
|
| 172 |
createHeader(writer, firstLine); |
|
| 173 |
|
|
| 174 |
//write data |
|
| 175 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 176 |
|
|
| 177 |
if (!merging.getRankInErms().equals(merging.getRankInFaunaEu())){
|
|
| 178 |
writeCsvLine(writer, merging); |
|
| 179 |
} |
|
| 180 |
} |
|
| 181 |
writer.flush(); |
|
| 182 |
writer.close(); |
|
| 183 |
return true; |
|
| 184 |
}catch(IOException e){
|
|
| 185 |
return false; |
|
| 186 |
} |
|
| 187 |
} |
|
| 188 |
|
|
| 189 |
private void createHeader(FileWriter writer, String firstLine) throws IOException{
|
|
| 190 |
writer.append(firstLine); |
|
| 191 |
writer.append('\n');
|
|
| 192 |
writer.append("uuid in Fauna Europaea");
|
|
| 193 |
writer.append(';');
|
|
| 194 |
writer.append("id in Fauna Europaea");
|
|
| 195 |
writer.append(';');
|
|
| 196 |
writer.append("name");
|
|
| 197 |
writer.append(';');
|
|
| 198 |
writer.append("author");
|
|
| 199 |
writer.append(';');
|
|
| 200 |
writer.append("rank");
|
|
| 201 |
writer.append(';');
|
|
| 202 |
writer.append("state");
|
|
| 203 |
writer.append(';');
|
|
| 204 |
writer.append("phylum");
|
|
| 205 |
writer.append(';');
|
|
| 206 |
writer.append("parent");
|
|
| 207 |
writer.append(';');
|
|
| 208 |
writer.append("parent rank");
|
|
| 209 |
writer.append(';');
|
|
| 210 |
|
|
| 211 |
writer.append("uuid in Erms");
|
|
| 212 |
writer.append(';');
|
|
| 213 |
writer.append("id in Erms");
|
|
| 214 |
writer.append(';');
|
|
| 215 |
writer.append("name");
|
|
| 216 |
writer.append(';');
|
|
| 217 |
writer.append("author");
|
|
| 218 |
writer.append(';');
|
|
| 219 |
writer.append("rank");
|
|
| 220 |
writer.append(';');
|
|
| 221 |
writer.append("state");
|
|
| 222 |
writer.append(';');
|
|
| 223 |
writer.append("phylum");
|
|
| 224 |
writer.append(';');
|
|
| 225 |
writer.append("parent");
|
|
| 226 |
writer.append(';');
|
|
| 227 |
writer.append("parent rank");
|
|
| 228 |
writer.append('\n');
|
|
| 229 |
} |
|
| 230 |
|
|
| 231 |
private boolean writeSameNamesDifferentStatusToCsv( |
|
| 232 |
List<PesiMergeObject> mergingObjects, String sfileName){
|
|
| 233 |
|
|
| 234 |
try{
|
|
| 235 |
FileWriter writer = new FileWriter(sfileName); |
|
| 236 |
|
|
| 237 |
//create Header |
|
| 238 |
String firstLine = "same names but different status"; |
|
| 239 |
createHeader(writer, firstLine); |
|
| 240 |
|
|
| 241 |
//write data |
|
| 242 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 243 |
if (merging.isStatInErms()^merging.isStatInFaunaEu()){
|
|
| 244 |
writeCsvLine(writer, merging); |
|
| 245 |
} |
|
| 246 |
} |
|
| 247 |
|
|
| 248 |
writer.flush(); |
|
| 249 |
writer.close(); |
|
| 250 |
return true; |
|
| 251 |
}catch(IOException e){
|
|
| 252 |
return false; |
|
| 253 |
} |
|
| 254 |
} |
|
| 255 |
|
|
| 256 |
private boolean writeSameNamesDifferentAuthorToCsv( |
|
| 257 |
List<PesiMergeObject> mergingObjects, String sfileName){
|
|
| 258 |
|
|
| 259 |
try{
|
|
| 260 |
FileWriter writer = new FileWriter(sfileName); |
|
| 261 |
|
|
| 262 |
//create Header |
|
| 263 |
String firstLine = "same names but different authors"; |
|
| 264 |
createHeader(writer, firstLine); |
|
| 265 |
|
|
| 266 |
//write data |
|
| 267 |
for (PesiMergeObject merging : mergingObjects){
|
|
| 268 |
|
|
| 269 |
if (merging.getAuthorInErms() != null && merging.getAuthorInFaunaEu() != null && !merging.getAuthorInErms().equals(merging.getAuthorInFaunaEu())){
|
|
| 270 |
writeCsvLine(writer, merging); |
|
| 271 |
}else if ((merging.getAuthorInErms() == null && merging.getAuthorInFaunaEu() != null) || (merging.getAuthorInErms() != null && merging.getAuthorInFaunaEu() == null)) {
|
|
| 272 |
writeCsvLine(writer, merging); |
|
| 273 |
} |
|
| 274 |
} |
|
| 275 |
|
|
| 276 |
writer.flush(); |
|
| 277 |
writer.close(); |
|
| 278 |
return true; |
|
| 279 |
}catch(IOException e){
|
|
| 280 |
return false; |
|
| 281 |
} |
|
| 282 |
} |
|
| 283 |
|
|
| 284 |
private void writeCsvLine(FileWriter writer, PesiMergeObject merging) throws IOException{
|
|
| 285 |
|
|
| 286 |
writer.append(merging.getUuidFaunaEu()); |
|
| 287 |
writer.append(';');
|
|
| 288 |
writer.append(merging.getIdInFaunaEu()); |
|
| 289 |
writer.append(';');
|
|
| 290 |
writer.append(merging.getNameCacheInFaunaEu()); |
|
| 291 |
writer.append(';');
|
|
| 292 |
writer.append(merging.getAuthorInFaunaEu()); |
|
| 293 |
writer.append(';');
|
|
| 294 |
writer.append(merging.getRankInFaunaEu()); |
|
| 295 |
writer.append(';');
|
|
| 296 |
if (merging.isStatInFaunaEu()){
|
|
| 297 |
writer.append("accepted");
|
|
| 298 |
}else{
|
|
| 299 |
writer.append("synonym");
|
|
| 300 |
} |
|
| 301 |
writer.append(';');
|
|
| 302 |
writer.append(merging.getPhylumInFaunaEu() != null? merging.getPhylumInFaunaEu().getTaxonTitleCache(): ""); |
|
| 303 |
writer.append(';');
|
|
| 304 |
writer.append(merging.getParentStringInFaunaEu()); |
|
| 305 |
writer.append(';');
|
|
| 306 |
writer.append(merging.getParentRankStringInFaunaEu()); |
|
| 307 |
writer.append(';');
|
|
| 308 |
|
|
| 309 |
writer.append(merging.getUuidErms()); |
|
| 310 |
writer.append(';');
|
|
| 311 |
writer.append(merging.getIdInErms()); |
|
| 312 |
writer.append(';');
|
|
| 313 |
writer.append(merging.getNameCacheInErms()); |
|
| 314 |
writer.append(';');
|
|
| 315 |
writer.append(merging.getAuthorInErms()); |
|
| 316 |
writer.append(';');
|
|
| 317 |
writer.append(merging.getRankInErms()); |
|
| 318 |
writer.append(';');
|
|
| 319 |
if (merging.isStatInErms()){
|
|
| 320 |
writer.append("accepted");
|
|
| 321 |
}else{
|
|
| 322 |
writer.append("synonym");
|
|
| 323 |
} |
|
| 324 |
|
|
| 325 |
writer.append(';');
|
|
| 326 |
writer.append(merging.getPhylumInErms() != null? merging.getPhylumInErms().getTaxonTitleCache():""); |
|
| 327 |
writer.append(';');
|
|
| 328 |
writer.append(merging.getParentStringInErms()); |
|
| 329 |
writer.append(';');
|
|
| 330 |
writer.append(merging.getParentRankStringInErms()); |
|
| 331 |
writer.append('\n');
|
|
| 332 |
} |
|
| 333 |
|
|
| 334 |
private List<PesiMergeObject> createMergeObjects(Map<String,List<TaxonName>> names, |
|
| 335 |
CdmApplicationController appCtr){
|
|
| 336 |
|
|
| 337 |
UUID uuidClassification1 = UUID.fromString("44d8605e-a7ce-41e1-bee9-99edfec01e7c");
|
|
| 338 |
UUID uuidClassification2 = UUID.fromString("6fa988a9-10b7-48b0-a370-2586fbc066eb");
|
|
| 339 |
Classification classification1 = appCtr.getClassificationService().load(uuidClassification1); |
|
| 340 |
Classification classification2 = appCtr.getClassificationService().load(uuidClassification2); |
|
| 341 |
|
|
| 342 |
List<PesiMergeObject> merge = new ArrayList<>(); |
|
| 343 |
|
|
| 344 |
for (String nameCache: names.keySet()){
|
|
| 345 |
createSingleMergeObject(appCtr, merge, nameCache, names.get(nameCache),classification1, classification2); |
|
| 346 |
} |
|
| 347 |
|
|
| 348 |
return merge; |
|
| 349 |
} |
|
| 350 |
|
|
| 351 |
|
|
| 352 |
private void createSingleMergeObject(CdmApplicationController appCtr, List<PesiMergeObject> merge, String nameCache, |
|
| 353 |
List<TaxonName> identicalNames, |
|
| 354 |
Classification classification1, Classification classification2) {
|
|
| 355 |
|
|
| 356 |
PesiMergeObject mergeObject = new PesiMergeObject(); |
|
| 357 |
|
|
| 358 |
if(identicalNames.size()!= 2) {
|
|
| 359 |
logger.warn(nameCache + " has more than 2 names with identical name cache from different sources. This is not yet handled."); |
|
| 360 |
return; |
|
| 361 |
} |
|
| 362 |
//getPhylum |
|
| 363 |
TaxonNodeDto phylum1 = null; |
|
| 364 |
TaxonName faunaEuName = null; |
|
| 365 |
TaxonName ermsName = null; |
|
| 366 |
|
|
| 367 |
Set<TaxonBase> taxonBases = identicalNames.get(0).getTaxonBases(); |
|
| 368 |
if (taxonBases.size()==1) {
|
|
| 369 |
Iterator<TaxonBase> it = taxonBases.iterator(); |
|
| 370 |
TaxonBase<?> tempName = it.next(); |
|
| 371 |
if (tempName.getSec().getUuid().equals(faunaEuSecUuid)) {
|
|
| 372 |
faunaEuName = identicalNames.get(0); |
|
| 373 |
ermsName = identicalNames.get(1); |
|
| 374 |
}else {
|
|
| 375 |
faunaEuName = identicalNames.get(1); |
|
| 376 |
ermsName = identicalNames.get(0); |
|
| 377 |
} |
|
| 378 |
}else {
|
|
| 379 |
//TODO: find the two correct names |
|
| 380 |
logger.warn("Name has not exact 1 but " + taxonBases.size() + " taxon base attached. This is not yet handled");
|
|
| 381 |
return; |
|
| 382 |
} |
|
| 383 |
if (faunaEuName.getRank().equals(Rank.PHYLUM())) {
|
|
| 384 |
Taxon taxon = null; |
|
| 385 |
taxon = getAcceptedTaxon(faunaEuName); |
|
| 386 |
if (taxon != null) {
|
|
| 387 |
phylum1 = new TaxonNodeDto(taxon.getTaxonNode(classification1)); |
|
| 388 |
} |
|
| 389 |
|
|
| 390 |
} |
|
| 391 |
if (phylum1 == null && !faunaEuName.getRank().isHigher(Rank.PHYLUM())){
|
|
| 392 |
phylum1 = appCtr.getTaxonNodeService().taxonNodeDtoParentRank(classification1, Rank.PHYLUM(), faunaEuName); |
|
| 393 |
} |
|
| 394 |
|
|
| 395 |
TaxonNodeDto phylum2 = null; |
|
| 396 |
if (ermsName.getRank().equals(Rank.PHYLUM())) {
|
|
| 397 |
Taxon taxon = null; |
|
| 398 |
taxon = getAcceptedTaxon(ermsName); |
|
| 399 |
if (taxon != null) {
|
|
| 400 |
phylum2 = new TaxonNodeDto(taxon.getTaxonNode(classification2)); |
|
| 401 |
} |
|
| 402 |
|
|
| 403 |
} |
|
| 404 |
if (phylum2 == null && !ermsName.getRank().isHigher(Rank.PHYLUM())){
|
|
| 405 |
phylum2 = appCtr.getTaxonNodeService().taxonNodeDtoParentRank(classification2, Rank.PHYLUM(), ermsName); |
|
| 406 |
} |
|
| 407 |
mergeObject.setPhylumInErms(phylum1); |
|
| 408 |
mergeObject.setPhylumInFaunaEu(phylum2); |
|
| 409 |
|
|
| 410 |
//getUuids |
|
| 411 |
mergeObject.setUuidErms(ermsName.getUuid().toString()); |
|
| 412 |
mergeObject.setUuidFaunaEu(faunaEuName.getUuid().toString()); |
|
| 413 |
|
|
| 414 |
Iterator<IdentifiableSource> sources = ermsName.getSources().iterator(); |
|
| 415 |
if (sources.hasNext()){
|
|
| 416 |
IdentifiableSource source = sources.next(); |
|
| 417 |
String idInSource1 = source.getIdInSource(); |
|
| 418 |
mergeObject.setIdInErms(idInSource1); |
|
| 419 |
} |
|
| 420 |
sources = faunaEuName.getSources().iterator(); |
|
| 421 |
if (sources.hasNext()){
|
|
| 422 |
IdentifiableSource source = sources.next(); |
|
| 423 |
String idInSource1 = source.getIdInSource(); |
|
| 424 |
mergeObject.setIdInFaunaEu(idInSource1); |
|
| 425 |
} |
|
| 426 |
|
|
| 427 |
mergeObject.setNameCacheInErms(ermsName.getNameCache()); |
|
| 428 |
mergeObject.setNameCacheInFaunaEu(faunaEuName.getNameCache()); |
|
| 429 |
|
|
| 430 |
mergeObject.setAuthorInErms(ermsName.getAuthorshipCache()); |
|
| 431 |
mergeObject.setAuthorInFaunaEu(faunaEuName.getAuthorshipCache()); |
|
| 432 |
Set<Taxon> taxa = ermsName.getTaxa(); |
|
| 433 |
if (!taxa.isEmpty()){
|
|
| 434 |
mergeObject.setStatInErms(true); |
|
| 435 |
Iterator<Taxon> taxaIterator = taxa.iterator(); |
|
| 436 |
Taxon taxon = null; |
|
| 437 |
while (taxaIterator.hasNext()){
|
|
| 438 |
taxon = taxaIterator.next(); |
|
| 439 |
if (!taxon.isMisapplication()){
|
|
| 440 |
break; |
|
| 441 |
} |
|
| 442 |
} |
|
| 443 |
Set<TaxonNode> nodes = taxon.getTaxonNodes(); |
|
| 444 |
Iterator<TaxonNode> taxonNodeIterator = nodes.iterator(); |
|
| 445 |
TaxonNode node, parentNode = null; |
|
| 446 |
while (taxonNodeIterator.hasNext()){
|
|
| 447 |
node = taxonNodeIterator.next(); |
|
| 448 |
if (!node.isTopmostNode()){
|
|
| 449 |
parentNode = node.getParent(); |
|
| 450 |
} |
|
| 451 |
} |
|
| 452 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 453 |
if (parentNode != null){
|
|
| 454 |
TaxonName parentName = CdmBase.deproxy(parentNode.getTaxon().getName()); |
|
| 455 |
String parentNameCache = parentName.getNameCache(); |
|
| 456 |
mergeObject.setParentStringInErms(parentNameCache); |
|
| 457 |
mergeObject.setParentRankStringInErms(parentName.getRank().getLabel()); |
|
| 458 |
} |
|
| 459 |
}else{
|
|
| 460 |
mergeObject.setStatInErms(false); |
|
| 461 |
TaxonNode parentNode = getAcceptedNode(ermsName); |
|
| 462 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 463 |
if (parentNode != null){
|
|
| 464 |
TaxonName parentName = HibernateProxyHelper.deproxy(parentNode.getTaxon().getName()); |
|
| 465 |
String parentNameCache = parentName.getNameCache(); |
|
| 466 |
mergeObject.setParentStringInErms(parentNameCache); |
|
| 467 |
mergeObject.setParentRankStringInErms(parentName.getRank().getLabel()); |
|
| 468 |
} |
|
| 469 |
} |
|
| 470 |
taxa = faunaEuName.getTaxa(); |
|
| 471 |
if (!taxa.isEmpty()){
|
|
| 472 |
mergeObject.setStatInFaunaEu(true); |
|
| 473 |
Iterator<Taxon> taxaIterator = taxa.iterator(); |
|
| 474 |
Taxon taxon = null; |
|
| 475 |
while (taxaIterator.hasNext()){
|
|
| 476 |
taxon = taxaIterator.next(); |
|
| 477 |
if (!taxon.isMisapplication()){
|
|
| 478 |
break; |
|
| 479 |
} |
|
| 480 |
} |
|
| 481 |
Set<TaxonNode> nodes = taxon.getTaxonNodes(); |
|
| 482 |
Iterator<TaxonNode> taxonNodeIterator = nodes.iterator(); |
|
| 483 |
TaxonNode node, parentNode = null; |
|
| 484 |
while (taxonNodeIterator.hasNext()){
|
|
| 485 |
node = taxonNodeIterator.next(); |
|
| 486 |
if (!node.isTopmostNode()){
|
|
| 487 |
parentNode = node.getParent(); |
|
| 488 |
} |
|
| 489 |
} |
|
| 490 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 491 |
if (parentNode != null){
|
|
| 492 |
if (parentNode.getTaxon().getName().isZoological()){
|
|
| 493 |
|
|
| 494 |
IZoologicalName parentName = CdmBase.deproxy(parentNode.getTaxon().getName()); |
|
| 495 |
String parentNameCache = parentName.getNameCache(); |
|
| 496 |
mergeObject.setParentStringInFaunaEu(parentNameCache); |
|
| 497 |
mergeObject.setParentRankStringInFaunaEu(parentName.getRank().getLabel()); |
|
| 498 |
|
|
| 499 |
}else{
|
|
| 500 |
logger.debug("no zoologicalName: " + parentNode.getTaxon().getName().getTitleCache() +" . "+parentNode.getTaxon().getName().getUuid());
|
|
| 501 |
} |
|
| 502 |
} |
|
| 503 |
}else{
|
|
| 504 |
mergeObject.setStatInFaunaEu(false); |
|
| 505 |
TaxonNode parentNode = getAcceptedNode(faunaEuName); |
|
| 506 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 507 |
if (parentNode != null){
|
|
| 508 |
TaxonName parentName = HibernateProxyHelper.deproxy(parentNode.getTaxon().getName()); |
|
| 509 |
String parentNameCache = parentName.getNameCache(); |
|
| 510 |
mergeObject.setParentStringInFaunaEu(parentNameCache); |
|
| 511 |
mergeObject.setParentRankStringInFaunaEu(parentName.getRank().getLabel()); |
|
| 512 |
|
|
| 513 |
} |
|
| 514 |
} |
|
| 515 |
|
|
| 516 |
mergeObject.setRankInErms(ermsName.getRank().getLabel()); |
|
| 517 |
mergeObject.setRankInFaunaEu(faunaEuName.getRank().getLabel()); |
|
| 518 |
|
|
| 519 |
//set parent informations |
|
| 520 |
|
|
| 521 |
/* |
|
| 522 |
Set<HybridRelationship> parentRelations = zooName.getParentRelationships(); |
|
| 523 |
Iterator parentIterator = parentRelations.iterator(); |
|
| 524 |
HybridRelationship parentRel; |
|
| 525 |
ZoologicalName parentName; |
|
| 526 |
while (parentIterator.hasNext()){
|
|
| 527 |
parentRel = (HybridRelationship)parentIterator.next(); |
|
| 528 |
parentName = (ZoologicalName)parentRel.getParentName(); |
|
| 529 |
mergeObject.setParentRankStringInErms(parentName.getRank().getLabel()); |
|
| 530 |
mergeObject.setParentStringInErms(parentName.getNameCache()); |
|
| 531 |
} |
|
| 532 |
|
|
| 533 |
parentRelations = zooName2.getParentRelationships(); |
|
| 534 |
parentIterator = parentRelations.iterator(); |
|
| 535 |
|
|
| 536 |
while (parentIterator.hasNext()){
|
|
| 537 |
parentRel = (HybridRelationship)parentIterator.next(); |
|
| 538 |
parentName = (ZoologicalName)parentRel.getParentName(); |
|
| 539 |
mergeObject.setParentRankStringInFaunaEu(parentName.getRank().getLabel()); |
|
| 540 |
mergeObject.setParentStringInFaunaEu(parentName.getNameCache()); |
|
| 541 |
}*/ |
|
| 542 |
merge.add(mergeObject); |
|
| 543 |
} |
|
| 544 |
|
|
| 545 |
private TaxonNode getAcceptedNode(TaxonName ermsName) {
|
|
| 546 |
Set<TaxonBase> taxonBases = ermsName.getTaxonBases(); |
|
| 547 |
Taxon taxon = null; |
|
| 548 |
if (taxonBases != null && !taxonBases.isEmpty()) {
|
|
| 549 |
TaxonBase<?> taxonBase = taxonBases.iterator().next(); |
|
| 550 |
if (taxonBase instanceof Synonym) {
|
|
| 551 |
taxon = ((Synonym)taxonBase).getAcceptedTaxon(); |
|
| 552 |
} |
|
| 553 |
} |
|
| 554 |
|
|
| 555 |
Set<TaxonNode> nodes = taxon.getTaxonNodes(); |
|
| 556 |
|
|
| 557 |
TaxonNode parentNode = null; |
|
| 558 |
if (nodes != null && !nodes.isEmpty()) {
|
|
| 559 |
parentNode = nodes.iterator().next(); |
|
| 560 |
} |
|
| 561 |
return parentNode; |
|
| 562 |
} |
|
| 563 |
|
|
| 564 |
private Taxon getAcceptedTaxon(TaxonName ermsName) {
|
|
| 565 |
Taxon taxon = null; |
|
| 566 |
if (ermsName.getTaxa() != null && !ermsName.getTaxa().isEmpty()){
|
|
| 567 |
taxon = ermsName.getTaxa().iterator().next(); |
|
| 568 |
|
|
| 569 |
}else if (ermsName.getTaxonBases() != null && !ermsName.getTaxonBases().isEmpty()){
|
|
| 570 |
TaxonBase<?> taxonBase = ermsName.getTaxonBases().iterator().next(); |
|
| 571 |
if (taxonBase instanceof Synonym) {
|
|
| 572 |
Synonym syn = (Synonym)taxonBase; |
|
| 573 |
taxon = syn.getAcceptedTaxon(); |
|
| 574 |
} |
|
| 575 |
} |
|
| 576 |
return taxon; |
|
| 577 |
} |
|
| 578 |
|
|
| 579 |
public static void main(String[] args) {
|
|
| 580 |
FaunaEuErmsFindIdenticalNamesActivator activator = new FaunaEuErmsFindIdenticalNamesActivator(); |
|
| 581 |
activator.invoke(pesiSource); |
|
| 582 |
System.exit(0); |
|
| 583 |
} |
|
| 584 |
} |
|
| cdm-pesi/src/main/java/eu/etaxonomy/cdm/app/pesi/merging/PesiFindIdenticalNamesActivator.java | ||
|---|---|---|
| 1 |
package eu.etaxonomy.cdm.app.pesi.merging; |
|
| 2 |
|
|
| 3 |
import java.io.File; |
|
| 4 |
import java.io.FileWriter; |
|
| 5 |
import java.io.IOException; |
|
| 6 |
import java.lang.reflect.InvocationTargetException; |
|
| 7 |
import java.lang.reflect.Method; |
|
| 8 |
import java.util.ArrayList; |
|
| 9 |
import java.util.Arrays; |
|
| 10 |
import java.util.HashMap; |
|
| 11 |
import java.util.Iterator; |
|
| 12 |
import java.util.List; |
|
| 13 |
import java.util.Map; |
|
| 14 |
import java.util.Set; |
|
| 15 |
import java.util.UUID; |
|
| 16 |
|
|
| 17 |
import org.apache.log4j.Logger; |
|
| 18 |
|
|
| 19 |
import eu.etaxonomy.cdm.api.application.CdmApplicationController; |
|
| 20 |
import eu.etaxonomy.cdm.app.common.CdmDestinations; |
|
| 21 |
import eu.etaxonomy.cdm.common.CdmUtils; |
|
| 22 |
import eu.etaxonomy.cdm.common.StringComparator; |
|
| 23 |
import eu.etaxonomy.cdm.database.DbSchemaValidation; |
|
| 24 |
import eu.etaxonomy.cdm.database.ICdmDataSource; |
|
| 25 |
import eu.etaxonomy.cdm.io.api.application.CdmIoApplicationController; |
|
| 26 |
import eu.etaxonomy.cdm.io.pesi.merging.PesiMergeObject; |
|
| 27 |
import eu.etaxonomy.cdm.io.pesi.out.PesiTransformer; |
|
| 28 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 29 |
import eu.etaxonomy.cdm.model.common.IdentifiableSource; |
|
| 30 |
import eu.etaxonomy.cdm.model.name.Rank; |
|
| 31 |
import eu.etaxonomy.cdm.model.name.TaxonName; |
|
| 32 |
import eu.etaxonomy.cdm.model.taxon.Synonym; |
|
| 33 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
|
| 34 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
|
| 35 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode; |
|
| 36 |
import eu.etaxonomy.cdm.persistence.dto.TaxonNodeDto; |
|
| 37 |
|
|
| 38 |
public class PesiFindIdenticalNamesActivator {
|
|
| 39 |
|
|
| 40 |
private static final Logger logger = Logger.getLogger(PesiFindIdenticalNamesActivator.class); |
|
| 41 |
|
|
| 42 |
//static final ICdmDataSource faunaEuropaeaSource = CdmDestinations.localH2(); |
|
| 43 |
// static final ICdmDataSource pesiSource = CdmDestinations.cdm_test_local_faunaEu_mysql(); |
|
| 44 |
static final ICdmDataSource pesiSource = CdmDestinations.cdm_pesi2019_final(); |
|
| 45 |
|
|
| 46 |
static final String path = System.getProperty("user.home")+File.separator+".cdmLibrary"+File.separator+"pesi"+File.separator+"pesimerge";
|
|
| 47 |
|
|
| 48 |
private static UUID faunaEuSourceUuid = PesiTransformer.uuidSourceRefFaunaEuropaea; |
|
| 49 |
private static UUID ermsSourceUuid = PesiTransformer.uuidSourceRefErms; |
|
| 50 |
private static UUID ifSourceUuid = PesiTransformer.uuidSourceRefIndexFungorum; |
|
| 51 |
private static UUID emSourceUuid = PesiTransformer.uuidSourceRefEuroMed; |
|
| 52 |
private static List<UUID> sourceRefUuids = new ArrayList<>(); |
|
| 53 |
private static Map<UUID,String> sources = new HashMap<>(); |
|
| 54 |
|
|
| 55 |
static {
|
|
| 56 |
sourceRefUuids.addAll(Arrays.asList(new UUID[]{emSourceUuid, ermsSourceUuid, faunaEuSourceUuid, ifSourceUuid}));
|
|
| 57 |
sources.put(emSourceUuid, "E+M"); |
|
| 58 |
sources.put(ermsSourceUuid, "ERMS"); |
|
| 59 |
sources.put(faunaEuSourceUuid, "FauEu"); |
|
| 60 |
sources.put(ifSourceUuid, "IF"); |
|
| 61 |
} |
|
| 62 |
|
|
| 63 |
|
|
| 64 |
private void invoke(ICdmDataSource source){
|
|
| 65 |
|
|
| 66 |
CdmApplicationController app = CdmIoApplicationController.NewInstance(source, DbSchemaValidation.VALIDATE, false); |
|
| 67 |
|
|
| 68 |
List<String> propertyPaths = new ArrayList<>(); |
|
| 69 |
propertyPaths.add("sources.*");
|
|
| 70 |
propertyPaths.add("sources.idInSource");
|
|
| 71 |
propertyPaths.add("sources.idNamespace");
|
|
| 72 |
propertyPaths.add("taxonBases.*");
|
|
| 73 |
propertyPaths.add("taxonBases.relationsFromThisTaxon");
|
|
| 74 |
propertyPaths.add("taxonBases.taxonNodes.*");
|
|
| 75 |
propertyPaths.add("taxonBases.taxonNodes.parent.*");
|
|
| 76 |
propertyPaths.add("taxonBases.taxonNodes.childNodes.*");
|
|
| 77 |
propertyPaths.add("taxonBases.taxonNodes.childNodes.classification.rootNode.childNodes.*");
|
|
| 78 |
propertyPaths.add("taxonBases.taxonNodes.parent.taxon.name.*");
|
|
| 79 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.*");
|
|
| 80 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.childNodes.*");
|
|
| 81 |
propertyPaths.add("taxonBases.acceptedTaxon.taxonNodes.childNodes.classification.rootNode.childNodes.*");
|
|
| 82 |
System.out.println("Start getIdenticalNames...");
|
|
| 83 |
|
|
| 84 |
Map<String, Map<UUID, Set<TaxonName>>> namesOfIdenticalTaxa; |
|
| 85 |
try {
|
|
| 86 |
namesOfIdenticalTaxa = app.getTaxonService().findIdenticalTaxonNames(sourceRefUuids, propertyPaths); |
|
| 87 |
} catch (Exception e) {
|
|
| 88 |
e.printStackTrace(); |
|
| 89 |
return; |
|
| 90 |
} |
|
| 91 |
System.out.println("Start creating merging objects");
|
|
| 92 |
List<Map<UUID, PesiMergeObject>> mergingObjects = createMergeObjects(namesOfIdenticalTaxa, app); |
|
| 93 |
boolean resultOK = true; |
|
| 94 |
System.out.println("Start creating csv files");
|
|
| 95 |
resultOK &= writeSameNamesDifferentAuthorToCsv(mergingObjects, sources, path + "_authors.csv"); |
|
| 96 |
resultOK &= writeSameNamesDifferentStatusToCsv(mergingObjects, sources, path + "_status.csv"); |
|
| 97 |
resultOK &= writeSameNamesToCsvFile(mergingObjects, sources, path + "_names.csv"); |
|
| 98 |
resultOK &= writeSameNamesDifferentPhylumToCsv(mergingObjects, sources, path + "_phylum.csv"); |
|
| 99 |
resultOK &= writeSameNamesDifferentParentToCsv(mergingObjects, sources, path + "parent.csv"); |
|
| 100 |
resultOK &= writeSameNamesDifferentRankToCsv(mergingObjects, sources, path + "_rank.csv"); |
|
| 101 |
|
|
| 102 |
System.out.println("End find identical names for PESI: " + resultOK + ". Results written to " + path);
|
|
| 103 |
} |
|
| 104 |
|
|
| 105 |
private boolean writeSameNamesToCsvFile( |
|
| 106 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName) {
|
|
| 107 |
|
|
| 108 |
String header = "same names (all)"; |
|
| 109 |
String methodName = null; |
|
| 110 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 111 |
} |
|
| 112 |
|
|
| 113 |
private boolean writeSameNamesDifferentPhylumToCsv( |
|
| 114 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName){
|
|
| 115 |
|
|
| 116 |
String header = "same names but different phylum"; |
|
| 117 |
String methodName = "getPhylum"; |
|
| 118 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 119 |
} |
|
| 120 |
|
|
| 121 |
private boolean writeSameNamesDifferentParentToCsv( |
|
| 122 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName){
|
|
| 123 |
|
|
| 124 |
String header = "same names but different parent"; |
|
| 125 |
String methodName = "getParentString"; |
|
| 126 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 127 |
} |
|
| 128 |
|
|
| 129 |
private boolean writeSameNamesDifferentRankToCsv( |
|
| 130 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName){
|
|
| 131 |
|
|
| 132 |
String header = "same names but different rank"; |
|
| 133 |
String methodName = "getRank"; |
|
| 134 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 135 |
} |
|
| 136 |
|
|
| 137 |
private boolean writeSameNamesDifferentStatusToCsv( |
|
| 138 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName){
|
|
| 139 |
|
|
| 140 |
String header = "same names but different status"; |
|
| 141 |
String methodName = "isStatus"; |
|
| 142 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 143 |
} |
|
| 144 |
|
|
| 145 |
private boolean writeSameNamesDifferentAuthorToCsv( |
|
| 146 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName){
|
|
| 147 |
|
|
| 148 |
String header = "same names but different author"; |
|
| 149 |
String methodName = "getAuthor"; |
|
| 150 |
return writeDifference(header, methodName, mergingObjects, sources, sFileName); |
|
| 151 |
} |
|
| 152 |
|
|
| 153 |
private boolean writeDifference(String header, String methodName, |
|
| 154 |
List<Map<UUID, PesiMergeObject>> mergingObjects, Map<UUID,String> sources, String sFileName) {
|
|
| 155 |
|
|
| 156 |
try{
|
|
| 157 |
Method method = methodName == null? null : PesiMergeObject.class.getMethod(methodName); |
|
| 158 |
|
|
| 159 |
FileWriter writer = new FileWriter(sFileName); |
|
| 160 |
|
|
| 161 |
//create Header |
|
| 162 |
createHeader(writer, header); |
|
| 163 |
|
|
| 164 |
//write data |
|
| 165 |
for (Map<UUID, PesiMergeObject> merging : mergingObjects){
|
|
| 166 |
if (isDifferent(merging, method)){
|
|
| 167 |
writeCsvLine(writer, merging, sources) ; |
|
| 168 |
} |
|
| 169 |
} |
|
| 170 |
writer.flush(); |
|
| 171 |
writer.close(); |
|
| 172 |
return true; |
|
| 173 |
}catch(IOException | IllegalAccessException | IllegalArgumentException | InvocationTargetException | NoSuchMethodException | SecurityException e){
|
|
| 174 |
logger.error(e.getMessage()); |
|
| 175 |
return false; |
|
| 176 |
} |
|
| 177 |
} |
|
| 178 |
|
|
| 179 |
private boolean isDifferent(Map<UUID, PesiMergeObject> merging, Method method) throws IllegalAccessException, IllegalArgumentException, InvocationTargetException {
|
|
| 180 |
|
|
| 181 |
if (method == null){
|
|
| 182 |
return true; |
|
| 183 |
} |
|
| 184 |
Object value = null; |
|
| 185 |
boolean isFirst = true; |
|
| 186 |
for (UUID sourceUuid: merging.keySet()){
|
|
| 187 |
if (isFirst){
|
|
| 188 |
value = method.invoke(merging.get(sourceUuid)); |
|
| 189 |
isFirst = false; |
|
| 190 |
}else{
|
|
| 191 |
Object newValue = method.invoke(merging.get(sourceUuid)); |
|
| 192 |
if (!CdmUtils.nullSafeEqual(newValue, value)){
|
|
| 193 |
return true; |
|
| 194 |
} |
|
| 195 |
} |
|
| 196 |
} |
|
| 197 |
return false; |
|
| 198 |
} |
|
| 199 |
|
|
| 200 |
private void createHeader(FileWriter writer, String firstLine) throws IOException{
|
|
| 201 |
writer.append(firstLine); |
|
| 202 |
writer.append('\n');
|
|
| 203 |
|
|
| 204 |
for (int i=1; i<=2; i++){
|
|
| 205 |
writer.append("source"+i);
|
|
| 206 |
writer.append(';');
|
|
| 207 |
writer.append("name uuid"+i);
|
|
| 208 |
writer.append(';');
|
|
| 209 |
writer.append("name id"+i);
|
|
| 210 |
writer.append(';');
|
|
| 211 |
writer.append("name"+i);
|
|
| 212 |
writer.append(';');
|
|
| 213 |
writer.append("author"+i);
|
|
| 214 |
writer.append(';');
|
|
| 215 |
writer.append("rank"+i);
|
|
| 216 |
writer.append(';');
|
|
| 217 |
writer.append("state"+i);
|
|
| 218 |
writer.append(';');
|
|
| 219 |
writer.append("phylum"+i);
|
|
| 220 |
writer.append(';');
|
|
| 221 |
writer.append("parent"+i);
|
|
| 222 |
writer.append(';');
|
|
| 223 |
writer.append("parent rank"+i);
|
|
| 224 |
writer.append(';');
|
|
| 225 |
} |
|
| 226 |
writer.append('\n');
|
|
| 227 |
} |
|
| 228 |
|
|
| 229 |
private void writeCsvLine(FileWriter writer, Map<UUID,PesiMergeObject> mergeObjects, Map<UUID,String> sources) throws IOException{
|
|
| 230 |
|
|
| 231 |
for (UUID uuid : sourceRefUuids){
|
|
| 232 |
PesiMergeObject merging = mergeObjects.get(uuid); |
|
| 233 |
if(merging == null){
|
|
| 234 |
continue; |
|
| 235 |
} |
|
| 236 |
writer.append(sources.get(uuid)).append(";");
|
|
| 237 |
writer.append(merging.getUuidName()).append(";");
|
|
| 238 |
writer.append(merging.getIdInSource()).append(";");
|
|
| 239 |
writer.append(merging.getNameCache()).append(";");
|
|
| 240 |
writer.append(merging.getAuthor()).append(";");
|
|
| 241 |
writer.append(merging.getRank()).append(";");
|
|
| 242 |
if (merging.isStatus()){
|
|
| 243 |
writer.append("accepted").append(";");;
|
|
| 244 |
}else{
|
|
| 245 |
writer.append("synonym").append(";");;
|
|
| 246 |
} |
|
| 247 |
writer.append(merging.getPhylum() != null? merging.getPhylum().getTitleCache(): "").append(";");
|
|
| 248 |
writer.append(merging.getParentString()).append(";");
|
|
| 249 |
writer.append(merging.getParentRankString()).append(";");;
|
|
| 250 |
} |
|
| 251 |
writer.append('\n');
|
|
| 252 |
} |
|
| 253 |
|
|
| 254 |
private List<Map<UUID,PesiMergeObject>> createMergeObjects(Map<String, Map<UUID, Set<TaxonName>>> names, |
|
| 255 |
CdmApplicationController appCtr){
|
|
| 256 |
|
|
| 257 |
List<Map<UUID,PesiMergeObject>> merge = new ArrayList<>(); |
|
| 258 |
|
|
| 259 |
List<String> nameCaches = new ArrayList<>(names.keySet()); |
|
| 260 |
nameCaches.sort(StringComparator.Instance); |
|
| 261 |
for (String nameCache: nameCaches){
|
|
| 262 |
createSingleMergeObject(appCtr, merge, names.get(nameCache)); |
|
| 263 |
} |
|
| 264 |
|
|
| 265 |
return merge; |
|
| 266 |
} |
|
| 267 |
|
|
| 268 |
|
|
| 269 |
private void createSingleMergeObject(CdmApplicationController appCtr, List<Map<UUID,PesiMergeObject>> merge, |
|
| 270 |
Map<UUID, Set<TaxonName>> identicalNames) {
|
|
| 271 |
|
|
| 272 |
Map<UUID,PesiMergeObject> mergeMap = new HashMap<>(); |
|
| 273 |
|
|
| 274 |
for (UUID sourceUuid : identicalNames.keySet()){
|
|
| 275 |
Set<TaxonName> names = identicalNames.get(sourceUuid); |
|
| 276 |
if (names.isEmpty()){
|
|
| 277 |
continue; |
|
| 278 |
} |
|
| 279 |
TaxonName name = names.iterator().next(); |
|
| 280 |
String nameAndIdStr = name.getTitleCache() + "; id = " + name.getId(); |
|
| 281 |
if (names.size()>1){
|
|
| 282 |
logger.warn("Multiple names per source not yet handled. Take arbitrary one: " + nameAndIdStr);
|
|
| 283 |
} |
|
| 284 |
|
|
| 285 |
PesiMergeObject mergeObject = new PesiMergeObject(); |
|
| 286 |
mergeMap.put(sourceUuid, mergeObject); |
|
| 287 |
|
|
| 288 |
Set<TaxonBase> taxonBases = name.getTaxonBases(); |
|
| 289 |
if (taxonBases.isEmpty()){
|
|
| 290 |
logger.warn("No taxonbase attached to name. This is not yet handled: " + nameAndIdStr);
|
|
| 291 |
continue; |
|
| 292 |
} |
|
| 293 |
if (taxonBases.size() > 1) {
|
|
| 294 |
//TODO: find the two correct names |
|
| 295 |
logger.warn("Name has not exact 1 but " + taxonBases.size() + " taxon base attached. This is not yet handled. Take arbitrary one.");
|
|
| 296 |
} |
|
| 297 |
TaxonBase<?> taxonBase = taxonBases.iterator().next(); |
|
| 298 |
|
|
| 299 |
//uuid |
|
| 300 |
mergeObject.setUuidName(name.getUuid().toString()); |
|
| 301 |
|
|
| 302 |
//nameCache |
|
| 303 |
mergeObject.setNameCache(name.getNameCache()); |
|
| 304 |
|
|
| 305 |
//authorship |
|
| 306 |
mergeObject.setAuthor(name.getAuthorshipCache()); |
|
| 307 |
|
|
| 308 |
//rank |
|
| 309 |
mergeObject.setRank(name.getRank().getLabel()); |
|
| 310 |
|
|
| 311 |
//Phylum |
|
| 312 |
TaxonNodeDto phylum = null; |
|
| 313 |
if (name.getRank().equals(Rank.PHYLUM())) {
|
|
| 314 |
Taxon taxon = getAcceptedTaxon(name); |
|
| 315 |
if (taxon != null) {
|
|
| 316 |
if (taxon.getTaxonNodes().size()>1){
|
|
| 317 |
logger.warn("More than 1 node not yet handled for getPhylum. Take arbitrary one.");
|
|
| 318 |
} |
|
| 319 |
TaxonNode node = taxon.getTaxonNodes().iterator().next(); |
|
| 320 |
phylum = new TaxonNodeDto(node); |
|
| 321 |
} |
|
| 322 |
|
|
| 323 |
} |
|
| 324 |
if (phylum == null && !name.getRank().isHigher(Rank.PHYLUM())){
|
|
| 325 |
Taxon taxon = getAcceptedTaxon(name); |
|
| 326 |
if (!taxon.getTaxonNodes().isEmpty()){
|
|
| 327 |
if (taxon.getTaxonNodes().size()>1){
|
|
| 328 |
logger.warn("More than 1 node not yet handled for getPhylum. Take arbitrary one.");
|
|
| 329 |
} |
|
| 330 |
TaxonNode node = taxon.getTaxonNodes().iterator().next(); |
|
| 331 |
phylum = appCtr.getTaxonNodeService().taxonNodeDtoParentRank(node.getClassification(), Rank.PHYLUM(), name); |
|
| 332 |
} |
|
| 333 |
} |
|
| 334 |
mergeObject.setPhylum(phylum); |
|
| 335 |
|
|
| 336 |
//idInSource |
|
| 337 |
Iterator<IdentifiableSource> sources = name.getSources().iterator(); |
|
| 338 |
//TODO idInSource - what if multiple sources exist? |
|
| 339 |
if (sources.hasNext()){
|
|
| 340 |
IdentifiableSource source = sources.next(); |
|
| 341 |
String idInSource = source.getIdInSource(); |
|
| 342 |
mergeObject.setIdInSource(idInSource); |
|
| 343 |
} |
|
| 344 |
|
|
| 345 |
//status and parent |
|
| 346 |
Set<Taxon> taxa = name.getTaxa(); |
|
| 347 |
if (!taxa.isEmpty()){
|
|
| 348 |
mergeObject.setStatus(true); |
|
| 349 |
Iterator<Taxon> taxaIterator = taxa.iterator(); |
|
| 350 |
Taxon taxon = null; |
|
| 351 |
while (taxaIterator.hasNext()){
|
|
| 352 |
taxon = taxaIterator.next(); |
|
| 353 |
if (!taxon.isMisapplication()){
|
|
| 354 |
break; |
|
| 355 |
} |
|
| 356 |
} |
|
| 357 |
@SuppressWarnings("null")
|
|
| 358 |
Set<TaxonNode> nodes = taxon.getTaxonNodes(); |
|
| 359 |
Iterator<TaxonNode> taxonNodeIterator = nodes.iterator(); |
|
| 360 |
TaxonNode parentNode = null; |
|
| 361 |
while (taxonNodeIterator.hasNext()){
|
|
| 362 |
TaxonNode node = taxonNodeIterator.next(); |
|
| 363 |
if (!node.isTopmostNode()){
|
|
| 364 |
parentNode = node.getParent(); |
|
| 365 |
} |
|
| 366 |
} |
|
| 367 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 368 |
if (parentNode != null){
|
|
| 369 |
TaxonName parentName = CdmBase.deproxy(parentNode.getTaxon().getName()); |
|
| 370 |
String parentNameCache = parentName.getNameCache(); |
|
| 371 |
mergeObject.setParentString(parentNameCache); |
|
| 372 |
mergeObject.setParentRankString(parentName.getRank().getLabel()); |
|
| 373 |
} |
|
| 374 |
}else{
|
|
| 375 |
mergeObject.setStatus(false); |
|
| 376 |
TaxonNode parentNode = getAcceptedNode(name); |
|
| 377 |
//TODO: ändern mit erweitertem Initializer.. |
|
| 378 |
if (parentNode != null){
|
|
| 379 |
TaxonName parentName = CdmBase.deproxy(parentNode.getTaxon().getName()); |
|
| 380 |
String parentNameCache = parentName.getNameCache(); |
|
| 381 |
mergeObject.setParentString(parentNameCache); |
|
| 382 |
mergeObject.setParentRankString(parentName.getRank().getLabel()); |
|
| 383 |
} |
|
| 384 |
} |
|
| 385 |
} |
|
| 386 |
|
|
| 387 |
|
|
| 388 |
//set parent informations |
|
| 389 |
|
|
| 390 |
/* |
|
| 391 |
Set<HybridRelationship> parentRelations = zooName.getParentRelationships(); |
|
| 392 |
Iterator parentIterator = parentRelations.iterator(); |
|
| 393 |
HybridRelationship parentRel; |
|
| 394 |
ZoologicalName parentName; |
|
| 395 |
while (parentIterator.hasNext()){
|
|
| 396 |
parentRel = (HybridRelationship)parentIterator.next(); |
|
| 397 |
parentName = (ZoologicalName)parentRel.getParentName(); |
|
| 398 |
mergeObject.setParentRankStringInErms(parentName.getRank().getLabel()); |
|
| 399 |
mergeObject.setParentStringInErms(parentName.getNameCache()); |
|
| 400 |
} |
|
| 401 |
|
|
| 402 |
parentRelations = zooName2.getParentRelationships(); |
|
| 403 |
parentIterator = parentRelations.iterator(); |
|
| 404 |
|
|
| 405 |
while (parentIterator.hasNext()){
|
|
| 406 |
parentRel = (HybridRelationship)parentIterator.next(); |
|
| 407 |
parentName = (ZoologicalName)parentRel.getParentName(); |
|
| 408 |
mergeObject.setParentRankStringInFaunaEu(parentName.getRank().getLabel()); |
|
| 409 |
mergeObject.setParentStringInFaunaEu(parentName.getNameCache()); |
|
| 410 |
}*/ |
|
| 411 |
|
|
| 412 |
merge.add(mergeMap); |
|
| 413 |
} |
|
| 414 |
|
|
| 415 |
private TaxonNode getAcceptedNode(TaxonName ermsName) {
|
|
| 416 |
Set<TaxonBase> taxonBases = ermsName.getTaxonBases(); |
|
| 417 |
Taxon taxon = null; |
|
| 418 |
if (taxonBases != null && !taxonBases.isEmpty()) {
|
|
| 419 |
TaxonBase<?> taxonBase = taxonBases.iterator().next(); |
|
| 420 |
if (taxonBase instanceof Synonym) {
|
|
| 421 |
taxon = ((Synonym)taxonBase).getAcceptedTaxon(); |
|
| 422 |
} |
|
| 423 |
} |
|
| 424 |
|
|
| 425 |
Set<TaxonNode> nodes = taxon.getTaxonNodes(); |
|
| 426 |
|
|
| 427 |
TaxonNode parentNode = null; |
|
| 428 |
if (nodes != null && !nodes.isEmpty()) {
|
|
| 429 |
parentNode = nodes.iterator().next(); |
|
| 430 |
} |
|
| 431 |
return parentNode; |
|
| 432 |
} |
|
| 433 |
|
|
| 434 |
private Taxon getAcceptedTaxon(TaxonName name) {
|
|
| 435 |
Taxon taxon = null; |
|
| 436 |
//prefer accepted taxon |
|
| 437 |
if (name.getTaxa() != null && !name.getTaxa().isEmpty()){
|
|
| 438 |
taxon = name.getTaxa().iterator().next(); |
|
| 439 |
//else take synonym |
|
| 440 |
}else if (name.getTaxonBases() != null && !name.getTaxonBases().isEmpty()){
|
|
| 441 |
TaxonBase<?> taxonBase = name.getTaxonBases().iterator().next(); |
|
| 442 |
if (taxonBase instanceof Synonym) {
|
|
| 443 |
Synonym syn = (Synonym)taxonBase; |
|
| 444 |
taxon = syn.getAcceptedTaxon(); |
|
| 445 |
} |
|
| 446 |
} |
|
| 447 |
return taxon; |
|
| 448 |
} |
|
| 449 |
|
|
| 450 |
public static void main(String[] args) {
|
|
| 451 |
PesiFindIdenticalNamesActivator activator = new PesiFindIdenticalNamesActivator(); |
|
| 452 |
activator.invoke(pesiSource); |
|
| 453 |
System.exit(0); |
|
| 454 |
} |
|
| 455 |
} |
|
| cdm-pesi/src/main/java/eu/etaxonomy/cdm/io/pesi/merging/PesiMergeObject.java | ||
|---|---|---|
| 4 | 4 |
|
| 5 | 5 |
public class PesiMergeObject {
|
| 6 | 6 |
|
| 7 |
private String uuidFaunaEu; |
|
| 8 |
private String uuidErms; |
|
| 7 |
private String uuidName; |
|
| 9 | 8 |
|
| 10 |
private String idInFaunaEu; |
|
| 11 |
private String idInErms; |
|
| 9 |
private String idInSource; |
|
| 12 | 10 |
|
| 13 |
private String nameCacheInFaunaEu; |
|
| 14 |
private String nameCacheInErms; |
|
| 11 |
private String nameCache; |
|
| 15 | 12 |
|
| 16 |
private boolean statInFaunaEu; |
|
| 17 |
private boolean statInErms; |
|
| 13 |
private boolean status; |
|
| 18 | 14 |
|
| 19 |
private String authorInFaunaEu; |
|
| 20 |
private String authorInErms; |
|
| 15 |
private String author; |
|
| 21 | 16 |
|
| 22 |
private String rankInFaunaEu; |
|
| 23 |
private String rankInErms; |
|
| 17 |
private String rank; |
|
| 24 | 18 |
|
| 25 |
private TaxonNodeDto phylumInFaunaEu; |
|
| 26 |
private TaxonNodeDto phylumInErms; |
|
| 19 |
private TaxonNodeDto phylum; |
|
| 27 | 20 |
|
| 28 |
private String parentStringInFaunaEu; |
|
| 29 |
private String parentStringInErms; |
|
| 21 |
private String parentString; |
|
| 30 | 22 |
|
| 31 |
private String parentRankStringInFaunaEu; |
|
| 32 |
private String parentRankStringInErms; |
|
| 23 |
private String parentRankString; |
|
| 33 | 24 |
|
| 34 | 25 |
|
| 35 |
public String getUuidFaunaEu() {
|
|
| 36 |
return uuidFaunaEu; |
|
| 37 |
} |
|
| 38 |
public void setUuidFaunaEu(String uuidFaunaEu) {
|
|
| 39 |
this.uuidFaunaEu = uuidFaunaEu; |
|
| 40 |
} |
|
| 26 |
//************************ FACTORY *******************/ |
|
| 41 | 27 |
|
| 42 |
public String getUuidErms() {
|
|
| 43 |
return uuidErms; |
|
| 44 |
} |
|
| 28 |
public static PesiMergeObject newInstance(){
|
|
| 29 |
return new PesiMergeObject(); |
|
| 45 | 30 |
|
| 46 |
public void setUuidErms(String uuidErms) {
|
|
| 47 |
this.uuidErms = uuidErms; |
|
| 48 |
} |
|
| 31 |
} |
|
| 49 | 32 |
|
| 50 |
public String getParentRankStringInFaunaEu() {
|
|
| 51 |
return parentRankStringInFaunaEu; |
|
| 52 |
} |
|
| 33 |
//************************* GETTER/SETTER **********************/ |
|
| 53 | 34 |
|
| 54 |
public void setParentRankStringInFaunaEu(String parentRankStringInFaunaEu) {
|
|
| 55 |
this.parentRankStringInFaunaEu = parentRankStringInFaunaEu;
|
|
| 35 |
public String getUuidName() {
|
|
| 36 |
return uuidName;
|
|
| 56 | 37 |
} |
| 57 |
|
|
| 58 |
public String getParentRankStringInErms() {
|
|
| 59 |
return parentRankStringInErms; |
|
| 38 |
public void setUuidName(String uuidName) {
|
|
| 39 |
this.uuidName = uuidName; |
|
| 60 | 40 |
} |
| 61 | 41 |
|
| 62 |
public void setParentRankStringInErms(String parentRankStringInErms) {
|
|
| 63 |
this.parentRankStringInErms = parentRankStringInErms;
|
|
| 42 |
public String getParentRankString() {
|
|
| 43 |
return parentRankString;
|
|
| 64 | 44 |
} |
| 65 |
|
|
| 66 |
public String getParentStringInFaunaEu() {
|
|
| 67 |
return parentStringInFaunaEu; |
|
| 45 |
public void setParentRankString(String parentRankString) {
|
|
| 46 |
this.parentRankString = parentRankString; |
|
| 68 | 47 |
} |
| 69 | 48 |
|
| 70 |
public void setParentStringInFaunaEu(String parentStringInFaunaEu) {
|
|
| 71 |
this.parentStringInFaunaEu = parentStringInFaunaEu;
|
|
| 49 |
public String getParentString() {
|
|
| 50 |
return parentString;
|
|
| 72 | 51 |
} |
| 73 |
|
|
| 74 |
public String getParentStringInErms() {
|
|
| 75 |
return parentStringInErms; |
|
| 52 |
public void setParentString(String parentString) {
|
|
| 53 |
this.parentString = parentString; |
|
| 76 | 54 |
} |
| 77 | 55 |
|
| 78 |
public void setParentStringInErms(String parentStringInErms) {
|
|
| 79 |
this.parentStringInErms = parentStringInErms;
|
|
| 56 |
public String getRank() {
|
|
| 57 |
return rank;
|
|
| 80 | 58 |
} |
| 81 |
|
|
| 82 |
public String getRankInFaunaEu() {
|
|
| 83 |
return rankInFaunaEu; |
|
| 59 |
public void setRank(String rank) {
|
|
| 60 |
this.rank = rank; |
|
| 84 | 61 |
} |
| 85 | 62 |
|
| 86 |
public void setRankInFaunaEu(String rankInFaunaEu) {
|
|
| 87 |
this.rankInFaunaEu = rankInFaunaEu;
|
|
| 63 |
public TaxonNodeDto getPhylum() {
|
|
| 64 |
return phylum;
|
|
| 88 | 65 |
} |
| 89 |
|
|
| 90 |
public TaxonNodeDto getPhylumInFaunaEu() {
|
|
| 91 |
return phylumInFaunaEu; |
|
| 66 |
public void setPhylum(TaxonNodeDto phylum) {
|
|
| 67 |
this.phylum = phylum; |
|
| 92 | 68 |
} |
| 93 | 69 |
|
| 94 |
public void setPhylumInFaunaEu(TaxonNodeDto phylumInFaunaEu) {
|
|
| 95 |
this.phylumInFaunaEu = phylumInFaunaEu;
|
|
| 70 |
public boolean isStatus() {
|
|
| 71 |
return status;
|
|
| 96 | 72 |
} |
| 97 |
|
|
| 98 |
public TaxonNodeDto getPhylumInErms() {
|
|
| 99 |
return phylumInErms; |
|
| 73 |
public void setStatus(boolean status) {
|
|
| 74 |
this.status = status; |
|
| 100 | 75 |
} |
| 101 | 76 |
|
| 102 |
public void setPhylumInErms(TaxonNodeDto phylumInErms) {
|
|
| 103 |
this.phylumInErms = phylumInErms;
|
|
| 77 |
public String getAuthor() {
|
|
| 78 |
return author;
|
|
| 104 | 79 |
} |
| 105 |
|
|
| 106 |
public String getRankInErms() {
|
|
| 107 |
return rankInErms; |
|
| 80 |
public void setAuthor(String author) {
|
|
| 81 |
this.author = author; |
|
| 108 | 82 |
} |
| 109 | 83 |
|
| 110 |
public void setRankInErms(String rankInErms) {
|
|
| 111 |
this.rankInErms = rankInErms;
|
|
| 84 |
public String getIdInSource() {
|
|
| 85 |
return idInSource;
|
|
| 112 | 86 |
} |
| 113 |
|
|
| 114 |
|
|
| 115 |
public static PesiMergeObject newInstance(){
|
|
| 116 |
return new PesiMergeObject(); |
|
| 117 |
|
|
| 87 |
public void setIdInSource(String idInSource) {
|
|
| 88 |
this.idInSource = idInSource; |
|
| 118 | 89 |
} |
| 119 | 90 |
|
| 120 |
public boolean isStatInFaunaEu() {
|
|
| 121 |
return statInFaunaEu; |
|
| 122 |
} |
|
| 123 |
public void setStatInFaunaEu(boolean statInFaunaEu) {
|
|
| 124 |
this.statInFaunaEu = statInFaunaEu; |
|
| 91 |
public String getNameCache() {
|
|
| 92 |
return nameCache; |
|
| 125 | 93 |
} |
| 126 |
public boolean isStatInErms() {
|
|
| 127 |
return statInErms;
|
|
| 94 |
public void setNameCache(String nameCache) {
|
|
| 95 |
this.nameCache = nameCache;
|
|
| 128 | 96 |
} |
| 129 |
public void setStatInErms(boolean statInErms) {
|
|
| 130 |
this.statInErms = statInErms; |
|
| 131 |
} |
|
| 132 |
public String getAuthorInFaunaEu() {
|
|
| 133 |
return authorInFaunaEu; |
|
| 134 |
} |
|
| 135 |
public void setAuthorInFaunaEu(String authorInFaunaEu) {
|
|
| 136 |
this.authorInFaunaEu = authorInFaunaEu; |
|
| 137 |
} |
|
| 138 |
public String getAuthorInErms() {
|
|
| 139 |
return authorInErms; |
|
| 140 |
} |
|
| 141 |
public void setAuthorInErms(String authorInErms) {
|
|
| 142 |
this.authorInErms = authorInErms; |
|
| 143 |
} |
|
| 144 |
public String getIdInFaunaEu() {
|
|
| 145 |
return idInFaunaEu; |
|
| 146 |
} |
|
| 147 |
public void setIdInFaunaEu(String idInFaunaEu) {
|
|
| 148 |
this.idInFaunaEu = idInFaunaEu; |
|
| 149 |
} |
|
| 150 |
public String getIdInErms() {
|
|
| 151 |
return idInErms; |
|
| 152 |
} |
|
| 153 |
public void setIdInErms(String idInErms) {
|
|
| 154 |
this.idInErms = idInErms; |
|
| 155 |
} |
|
| 156 |
public String getNameCacheInFaunaEu() {
|
|
| 157 |
return nameCacheInFaunaEu; |
|
| 158 |
} |
|
| 159 |
public void setNameCacheInFaunaEu(String nameCacheInFaunaEu) {
|
|
| 160 |
this.nameCacheInFaunaEu = nameCacheInFaunaEu; |
|
| 161 |
} |
|
| 162 |
public String getNameCacheInErms() {
|
|
| 163 |
return nameCacheInErms; |
|
| 164 |
} |
|
| 165 |
public void setNameCacheInErms(String nameCacheInErms) {
|
|
| 166 |
this.nameCacheInErms = nameCacheInErms; |
|
| 167 |
} |
|
| 168 |
|
|
| 169 |
|
|
| 170 |
|
|
| 171 | 97 |
} |
Also available in: Unified diff
ref #1447 improve PesiFindIdenticalNamesActivator