|
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.ArrayList;
|
|
12 |
import java.util.HashMap;
|
|
13 |
import java.util.HashSet;
|
|
14 |
import java.util.Iterator;
|
|
15 |
import java.util.List;
|
|
16 |
import java.util.Map;
|
|
17 |
import java.util.Set;
|
|
18 |
import java.util.UUID;
|
|
19 |
|
|
20 |
import org.apache.log4j.Logger;
|
|
21 |
import org.springframework.stereotype.Component;
|
|
22 |
|
|
23 |
import eu.etaxonomy.cdm.api.service.config.SynonymDeletionConfigurator;
|
|
24 |
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
25 |
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException;
|
|
26 |
import eu.etaxonomy.cdm.io.common.utils.ImportDeduplicationHelper;
|
|
27 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImport;
|
|
28 |
import eu.etaxonomy.cdm.io.mexico.SimpleExcelTaxonImportState;
|
|
29 |
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
30 |
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
|
31 |
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
|
32 |
import eu.etaxonomy.cdm.model.name.Rank;
|
|
33 |
import eu.etaxonomy.cdm.model.name.TaxonName;
|
|
34 |
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
35 |
import eu.etaxonomy.cdm.model.reference.ReferenceType;
|
|
36 |
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
|
37 |
import eu.etaxonomy.cdm.model.taxon.SynonymType;
|
|
38 |
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
|
39 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
40 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
|
41 |
import eu.etaxonomy.cdm.model.term.DefinedTerm;
|
|
42 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
|
43 |
|
|
44 |
/**
|
|
45 |
* @author a.mueller
|
|
46 |
* @since 17.02.2020
|
|
47 |
*/
|
|
48 |
@Component
|
|
49 |
public class CaryoAizoaceaeExcelImport extends SimpleExcelTaxonImport<CaryoAizoaceaeExcelImportConfigurator>{
|
|
50 |
|
|
51 |
private static final long serialVersionUID = -729761811965260921L;
|
|
52 |
private static final Logger logger = Logger.getLogger(CaryoAizoaceaeExcelImport.class);
|
|
53 |
|
|
54 |
private static final String ACCEPTED_PLANT_NAME_ID = "accepted_plant_name_id";
|
|
55 |
private static final String NOMENCLATURAL_REMARKS = "nomenclatural_remarks";
|
|
56 |
private static final String TAXON_RANK = "taxon_rank";
|
|
57 |
private static final String NAME_CIT = "NameCit";
|
|
58 |
private static final String KEW_NAME4CDM_LINK = "KewName4CDMLink";
|
|
59 |
private static final String KEW_F_NAME4CDM_LINK = "KewFName4CDMLink";
|
|
60 |
private static final String TAXON_STATUS = "taxon_status";
|
|
61 |
private static final String PLANT_NAME_ID = "plant_name_id";
|
|
62 |
private static final String IPNI_ID = "ipni_id";
|
|
63 |
|
|
64 |
private Map<String, UUID> taxonMapping = new HashMap<>();
|
|
65 |
private Reference secRef = null;
|
|
66 |
private Set<String> neglectedRecords = new HashSet<>();
|
|
67 |
private Set<UUID> createdNames = new HashSet<>();
|
|
68 |
|
|
69 |
private SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state;
|
|
70 |
private ImportDeduplicationHelper<SimpleExcelTaxonImportState<?>> dedupHelper = null;
|
|
71 |
|
|
72 |
@Override
|
|
73 |
protected void firstPass(SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state) {
|
|
74 |
int line = state.getCurrentLine();
|
|
75 |
if ((line % 500) == 0){
|
|
76 |
newTransaction(state);
|
|
77 |
System.out.println(line);
|
|
78 |
}
|
|
79 |
|
|
80 |
this.state = state;
|
|
81 |
Map<String, String> record = state.getOriginalRecord();
|
|
82 |
|
|
83 |
String fullCitation = getValue(record, NAME_CIT);
|
|
84 |
String nameCache = getValue(record, KEW_NAME4CDM_LINK);
|
|
85 |
String fullName = getValue(record, KEW_F_NAME4CDM_LINK);
|
|
86 |
String status = getValue(record, TAXON_STATUS);
|
|
87 |
String sourceId = getValue(record, PLANT_NAME_ID);
|
|
88 |
String ipniId = getValue(record, IPNI_ID);
|
|
89 |
String rankStr = getValue(record, TAXON_RANK);
|
|
90 |
String nomenclaturalRemarks = getValue(record, NOMENCLATURAL_REMARKS);
|
|
91 |
String accId = getValue(record, ACCEPTED_PLANT_NAME_ID);
|
|
92 |
|
|
93 |
String row = String.valueOf(line) + "("+fullName+"): ";
|
|
94 |
|
|
95 |
if("Misapplied".equals(status)){
|
|
96 |
neglectedRecords.add(sourceId);
|
|
97 |
return;
|
|
98 |
}
|
|
99 |
|
|
100 |
boolean isNewName = false;
|
|
101 |
|
|
102 |
try {
|
|
103 |
|
|
104 |
List<NomenclaturalStatusType> statusTypes = new ArrayList<>();
|
|
105 |
Class<? extends CdmBase> clazz = makeStatus(status, sourceId, accId, row, statusTypes);
|
|
106 |
|
|
107 |
TaxonName name;
|
|
108 |
Rank rank = state.getTransformer().getRankByKey(rankStr);
|
|
109 |
List<TaxonName> existingNames = getNameService().getNamesByNameCache(nameCache);
|
|
110 |
Iterator<TaxonName> it = existingNames.iterator();
|
|
111 |
while (it.hasNext()){
|
|
112 |
TaxonName next = it.next();
|
|
113 |
if (createdNames.contains(next.getUuid())){
|
|
114 |
it.remove();
|
|
115 |
}
|
|
116 |
}
|
|
117 |
|
|
118 |
List<TaxonName> fullNameMatches = new ArrayList<>();
|
|
119 |
|
|
120 |
@SuppressWarnings("rawtypes")
|
|
121 |
List<TaxonBase> allFullNameTaxa = new ArrayList<>();
|
|
122 |
@SuppressWarnings("rawtypes")
|
|
123 |
List<TaxonBase> allNameCacheTaxa = new ArrayList<>();
|
|
124 |
|
|
125 |
for (TaxonName existingName : existingNames){
|
|
126 |
if (existingName.getTitleCache().equals(fullName)){
|
|
127 |
fullNameMatches.add(existingName);
|
|
128 |
allFullNameTaxa.addAll(existingName.getTaxonBases());
|
|
129 |
}
|
|
130 |
allNameCacheTaxa.addAll(existingName.getTaxonBases());
|
|
131 |
}
|
|
132 |
|
|
133 |
logMultipleCandidates(row, existingNames, fullNameMatches);
|
|
134 |
|
|
135 |
TaxonBase<?> existingTaxon;
|
|
136 |
if(allFullNameTaxa.size()>1){
|
|
137 |
existingTaxon = findBestMatchingTaxon(allFullNameTaxa, clazz, row);
|
|
138 |
name = existingTaxon.getName();
|
|
139 |
}else if (allFullNameTaxa.size()==1){
|
|
140 |
existingTaxon = allFullNameTaxa.iterator().next();
|
|
141 |
name = existingTaxon.getName();
|
|
142 |
}else{
|
|
143 |
existingTaxon = null;
|
|
144 |
if (!fullNameMatches.isEmpty()){
|
|
145 |
logger.warn(row + "None of the existing names exists as taxon/synonym. Existing name taken as base for new taxon/synonym created.");
|
|
146 |
if (fullNameMatches.size()>1){
|
|
147 |
logger.warn(row + "More than 1 matching full names exist as candidats for new taxon/synonym. Arbitrary one taken.");
|
|
148 |
}
|
|
149 |
name = fullNameMatches.iterator().next();
|
|
150 |
}else if (!existingNames.isEmpty()){
|
|
151 |
if (!allNameCacheTaxa.isEmpty()){
|
|
152 |
logger.warn(row + "Taxa exist with matching nameCache but not matching fullname cache. New name and new taxon/synonym created. Other authors are " + getOtherAuthors(existingNames));
|
|
153 |
name = null;
|
|
154 |
}else{
|
|
155 |
logger.warn(row + "No matching fullnames exist but namecache matches. None of the matches is used in a taxon/synonym. Other authors are " + getOtherAuthors(existingNames));
|
|
156 |
name = null;
|
|
157 |
}
|
|
158 |
}else{
|
|
159 |
name = null;
|
|
160 |
}
|
|
161 |
}
|
|
162 |
|
|
163 |
if (existingTaxon == null){
|
|
164 |
if (rank == null){
|
|
165 |
logger.warn(row + "Name has no rank " + nameCache);
|
|
166 |
}else if (rank.equals(Rank.GENUS())){
|
|
167 |
logger.warn(row + "No name exists for genus " + nameCache + ". This is unexpected.");
|
|
168 |
}
|
|
169 |
}else{
|
|
170 |
if (existingTaxon.isInstanceOf(Taxon.class)){
|
|
171 |
if (!CdmBase.deproxy(existingTaxon, Taxon.class).getTaxonNodes().isEmpty()){
|
|
172 |
neglectedRecords.add(sourceId);
|
|
173 |
}
|
|
174 |
}else{
|
|
175 |
Taxon taxon = CdmBase.deproxy(existingTaxon, Synonym.class).getAcceptedTaxon();
|
|
176 |
if (taxon != null && !taxon.getTaxonNodes().isEmpty()){
|
|
177 |
neglectedRecords.add(sourceId);
|
|
178 |
}
|
|
179 |
}
|
|
180 |
}
|
|
181 |
if (name == null){
|
|
182 |
NonViralNameParserImpl parser = new NonViralNameParserImpl();
|
|
183 |
name = parser.parseReferencedName(fullCitation, NomenclaturalCode.ICNAFP, rank);
|
|
184 |
if (name.isProtectedFullTitleCache() || name.isProtectedTitleCache() || name.isProtectedNameCache()
|
|
185 |
|| name.isProtectedAuthorshipCache()){
|
|
186 |
logger.warn(row + "Name not parsable: " + fullCitation);
|
|
187 |
name.setTitleCache(fullName, true);
|
|
188 |
name.setNameCache(nameCache, true);
|
|
189 |
}else{
|
|
190 |
testParsedName(state, name, row, null);
|
|
191 |
}
|
|
192 |
name.addImportSource(sourceId, PLANT_NAME_ID, getSourceReference(state), "line " + state.getCurrentLine());
|
|
193 |
getNameService().saveOrUpdate(name);
|
|
194 |
isNewName = true;
|
|
195 |
createdNames.add(name.getUuid());
|
|
196 |
name = dedupliateNameParts(name);
|
|
197 |
}else{
|
|
198 |
testParsedName(state, name, row, fullCitation);
|
|
199 |
}
|
|
200 |
|
|
201 |
handleNomenclRemarkAndNameStatus(nomenclaturalRemarks, row, isNewName, name, statusTypes);
|
|
202 |
|
|
203 |
TaxonBase<?> taxonBase = existingTaxon;
|
|
204 |
|
|
205 |
if (taxonBase == null){
|
|
206 |
if (clazz == Taxon.class){
|
|
207 |
taxonBase = Taxon.NewInstance(name, getSecRef());
|
|
208 |
}else{
|
|
209 |
taxonBase = Synonym.NewInstance(name, getSecRef());
|
|
210 |
}
|
|
211 |
taxonBase.addImportSource(sourceId, PLANT_NAME_ID, getSourceReference(state), "line " + state.getCurrentLine());
|
|
212 |
getTaxonService().saveOrUpdate(taxonBase);
|
|
213 |
}
|
|
214 |
|
|
215 |
DefinedTerm ipniIdIdentifierType = DefinedTerm.IDENTIFIER_NAME_IPNI();
|
|
216 |
name.addIdentifier(ipniId, ipniIdIdentifierType);
|
|
217 |
|
|
218 |
taxonMapping.put(sourceId, taxonBase.getUuid());
|
|
219 |
// if("Accepted".equals(status)){
|
|
220 |
if(taxonBase.isInstanceOf(Taxon.class)){
|
|
221 |
UUID existingUuid = taxonMapping.put(name.getNameCache(), taxonBase.getUuid());
|
|
222 |
if (existingUuid != null){
|
|
223 |
logger.warn(row + name.getNameCache() + " has multiple instances in file");
|
|
224 |
}
|
|
225 |
}
|
|
226 |
} catch (UndefinedTransformerMethodException e) {
|
|
227 |
e.printStackTrace();
|
|
228 |
}
|
|
229 |
}
|
|
230 |
|
|
231 |
private TaxonName dedupliateNameParts(TaxonName name) {
|
|
232 |
getDedupHelper().replaceAuthorNamesAndNomRef(state, name);
|
|
233 |
return name;
|
|
234 |
}
|
|
235 |
|
|
236 |
private ImportDeduplicationHelper<SimpleExcelTaxonImportState<?>> getDedupHelper() {
|
|
237 |
if (dedupHelper == null){
|
|
238 |
dedupHelper
|
|
239 |
= ImportDeduplicationHelper.NewInstance(this, state);
|
|
240 |
}
|
|
241 |
return dedupHelper;
|
|
242 |
}
|
|
243 |
|
|
244 |
private String getOtherAuthors(List<TaxonName> otherNames) {
|
|
245 |
String result = "";
|
|
246 |
for (TaxonName name : otherNames){
|
|
247 |
result = CdmUtils.concat(";", result, name.getAuthorshipCache());
|
|
248 |
}
|
|
249 |
return result;
|
|
250 |
}
|
|
251 |
|
|
252 |
private TaxonBase<?> findBestMatchingTaxon(@SuppressWarnings("rawtypes") List<TaxonBase> allFullNameTaxa,
|
|
253 |
Class<? extends CdmBase> clazz, String row) {
|
|
254 |
|
|
255 |
TaxonBase<?> result = null;
|
|
256 |
TaxonBase<?> otherStatus = null;
|
|
257 |
for (TaxonBase<?> taxonBase : allFullNameTaxa) {
|
|
258 |
if (taxonBase.isInstanceOf(clazz)){
|
|
259 |
if (result != null){
|
|
260 |
logger.warn(row + "More than 1 taxon with matching full name AND matching status exists. This is not further handled. Arbitrary one taken.");
|
|
261 |
}
|
|
262 |
result = taxonBase;
|
|
263 |
}else{
|
|
264 |
otherStatus = taxonBase;
|
|
265 |
}
|
|
266 |
}
|
|
267 |
if (result == null && allFullNameTaxa.size()>1){
|
|
268 |
logger.warn(row + "More than 1 taxon with matching fullname but NOT matching status exists. This is not further handled. Arbitrary one taken.");
|
|
269 |
}
|
|
270 |
return result == null? otherStatus :result ;
|
|
271 |
}
|
|
272 |
|
|
273 |
private void logMultipleCandidates(String row, List<TaxonName> existingNames, List<TaxonName> fullNameMatches) {
|
|
274 |
if(fullNameMatches.size()>1){
|
|
275 |
String message = row + "More than one name with matching full name exists in DB. Try to take best matching.";
|
|
276 |
if (existingNames.size()>fullNameMatches.size()){
|
|
277 |
message += " Additionally names with matching name cache exist.";
|
|
278 |
}
|
|
279 |
logger.warn(message);
|
|
280 |
}else if (existingNames.size()>1){
|
|
281 |
String message = row + "More than one name with matching nameCache exists in DB. ";
|
|
282 |
if(fullNameMatches.isEmpty()){
|
|
283 |
message += "But none matches full name.";
|
|
284 |
}else{
|
|
285 |
message += "But exactly 1 matches full name.";
|
|
286 |
}
|
|
287 |
logger.warn(message);
|
|
288 |
}
|
|
289 |
}
|
|
290 |
|
|
291 |
private Class<? extends CdmBase> makeStatus(String status, String sourceId,
|
|
292 |
String accId, String row, List<NomenclaturalStatusType> statusTypes) {
|
|
293 |
|
|
294 |
Class<? extends CdmBase> clazz;
|
|
295 |
if ("Accepted".equals(status) || "Unplaced".equals(status) || "Artificial Hybrid".equals(status) ){
|
|
296 |
clazz = Taxon.class;
|
|
297 |
}else if ("Synonym".equals(status) || "Orthographic".equals(status)){
|
|
298 |
clazz = (accId == null)? Taxon.class : Synonym.class;
|
|
299 |
if("Orthographic".equals(status)){
|
|
300 |
statusTypes.add(NomenclaturalStatusType.SUPERFLUOUS());
|
|
301 |
// addStatus(NomenclaturalStatusType.SUPERFLUOUS(), row, isNewName, statusAdded, statusTypes, null);
|
|
302 |
}
|
|
303 |
}else if("Illegitimate".equals(status)){
|
|
304 |
clazz = getIllegInvalidStatus(sourceId, accId);
|
|
305 |
statusTypes.add(NomenclaturalStatusType.ILLEGITIMATE());
|
|
306 |
// addStatus(NomenclaturalStatusType.ILLEGITIMATE(), row, isNewName, statusAdded, statusTypes, getSecRef());
|
|
307 |
}else if ("Invalid".equals(status)){
|
|
308 |
clazz = getIllegInvalidStatus(sourceId, accId);
|
|
309 |
statusTypes.add(NomenclaturalStatusType.INVALID());
|
|
310 |
// addStatus(NomenclaturalStatusType.INVALID(), row, isNewName, statusAdded, statusTypes, getSecRef());
|
|
311 |
}else{
|
|
312 |
logger.warn(row + "Unhandled status: " + status);
|
|
313 |
clazz = Taxon.class; //to do something
|
|
314 |
}
|
|
315 |
return clazz;
|
|
316 |
}
|
|
317 |
|
|
318 |
private void handleNomenclRemarkAndNameStatus(String nomenclaturalRemarks, String row, boolean isNewName, TaxonName name,
|
|
319 |
List<NomenclaturalStatusType> statusTypes) {
|
|
320 |
|
|
321 |
NomenclaturalStatusType remarkType = null;
|
|
322 |
NomenclaturalStatusType statusType = statusTypes.isEmpty()? null: statusTypes.iterator().next();
|
|
323 |
if (nomenclaturalRemarks == null){
|
|
324 |
//nothing to do
|
|
325 |
}else if (", nom. illeg.".equals(nomenclaturalRemarks)){
|
|
326 |
remarkType = NomenclaturalStatusType.ILLEGITIMATE();
|
|
327 |
}else if (", nom. cons.".equals(nomenclaturalRemarks)){
|
|
328 |
remarkType = NomenclaturalStatusType.CONSERVED();
|
|
329 |
}else if (", nom. nud.".equals(nomenclaturalRemarks)){
|
|
330 |
remarkType = NomenclaturalStatusType.NUDUM();
|
|
331 |
}else if (", nom. provis.".equals(nomenclaturalRemarks)){
|
|
332 |
remarkType = NomenclaturalStatusType.PROVISIONAL();
|
|
333 |
}else if (", nom. rej.".equals(nomenclaturalRemarks)){
|
|
334 |
remarkType = NomenclaturalStatusType.REJECTED();
|
|
335 |
}else if (", nom. subnud.".equals(nomenclaturalRemarks)){
|
|
336 |
remarkType = NomenclaturalStatusType.SUBNUDUM();
|
|
337 |
}else if (", nom. superfl.".equals(nomenclaturalRemarks)){
|
|
338 |
remarkType = NomenclaturalStatusType.SUPERFLUOUS();
|
|
339 |
}else if (", not validly publ.".equals(nomenclaturalRemarks)){
|
|
340 |
statusTypes.add(NomenclaturalStatusType.INVALID());
|
|
341 |
}else if (", opus utique oppr.".equals(nomenclaturalRemarks)){
|
|
342 |
statusTypes.add(NomenclaturalStatusType.OPUS_UTIQUE_OPPR());
|
|
343 |
}else {
|
|
344 |
logger.warn(row + "Unhandled nomenclatural remark: " + nomenclaturalRemarks);
|
|
345 |
}
|
|
346 |
|
|
347 |
NomenclaturalStatusType kewType = remarkType != null? remarkType : statusType;
|
|
348 |
if (isNewName){
|
|
349 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
350 |
logger.warn(row + "Kew suggests 2 different nom. status. types for new name. The status from nomenclatural_remarks was taken.");
|
|
351 |
}
|
|
352 |
if (kewType != null){
|
|
353 |
name.addStatus(kewType, getSecRef(), null);
|
|
354 |
}
|
|
355 |
}else{
|
|
356 |
NomenclaturalStatusType existingType = null;
|
|
357 |
if (!name.getStatus().isEmpty()){
|
|
358 |
existingType = name.getStatus().iterator().next().getType();
|
|
359 |
}
|
|
360 |
if (existingType != null && kewType != null){
|
|
361 |
if (!existingType.equals(kewType)){
|
|
362 |
logger.warn(row + "Existing name status "+existingType.getTitleCache()+" differs from Kew status " + kewType.getTitleCache() + ". Key status ignored");
|
|
363 |
}
|
|
364 |
}else if (existingType != null && kewType == null){
|
|
365 |
logger.warn(row + "Info: Existing name has a name status "+existingType.getTitleCache()+" but Kew name has no status. Existing status kept.");
|
|
366 |
}else if (existingType == null && kewType != null){
|
|
367 |
if(remarkType != null && statusType != null && !remarkType.equals(statusType)){
|
|
368 |
logger.warn(row + "Existing name has no status while Kew name suggests a status (but 2 different status form status and nomenclatural_remarks field).");
|
|
369 |
}else{
|
|
370 |
logger.warn(row + "Existing name has no status while Kew name suggests a status ("+kewType.getTitleCache()+"). Kew status ignored.");
|
|
371 |
}
|
|
372 |
}
|
|
373 |
}
|
|
374 |
}
|
|
375 |
|
|
376 |
private void newTransaction(SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state) {
|
|
377 |
commitTransaction(state.getTransactionStatus());
|
|
378 |
secRef = null;
|
|
379 |
dedupHelper = null;
|
|
380 |
System.gc();
|
|
381 |
state.setTransactionStatus(startTransaction());
|
|
382 |
}
|
|
383 |
|
|
384 |
private Reference getSecRef() {
|
|
385 |
if (secRef == null){
|
|
386 |
secRef = getReferenceService().find(state.getConfig().getSecUuid());
|
|
387 |
}
|
|
388 |
return secRef;
|
|
389 |
}
|
|
390 |
|
|
391 |
private Class<? extends CdmBase> getIllegInvalidStatus(String sourceId, String accId) {
|
|
392 |
if (sourceId.equals(accId)){
|
|
393 |
return Taxon.class;
|
|
394 |
}else if(accId != null){
|
|
395 |
return Synonym.class;
|
|
396 |
}
|
|
397 |
return null;
|
|
398 |
}
|
|
399 |
|
|
400 |
|
|
401 |
private void testParsedName(SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state, TaxonName name,
|
|
402 |
String row, String fullCitation) throws UndefinedTransformerMethodException {
|
|
403 |
Map<String, String> record = state.getOriginalRecord();
|
|
404 |
|
|
405 |
String nameCache = getValue(record, KEW_NAME4CDM_LINK);
|
|
406 |
String fullName = getValue(record, KEW_F_NAME4CDM_LINK);
|
|
407 |
String rankStr = getValue(record, TAXON_RANK);
|
|
408 |
String genusHybrid = getValue(record, "genus_hybrid");
|
|
409 |
String genus = getValue(record, "genus");
|
|
410 |
String speciesHybrid = getValue(record, "species_hybrid");
|
|
411 |
String species = getValue(record, "species");
|
|
412 |
String infraSpecRank = getValue(record, "infraspecific_rank");
|
|
413 |
String infraspecies = getValue(record, "infraspecies");
|
|
414 |
String basionymAuthor = getValue(record, "parenthetical_author");
|
|
415 |
String combinationAuthor = getValue(record, "primary_author");
|
|
416 |
String authors = getValue(record, "taxon_authors");
|
|
417 |
String year = getValue(record, "KewYear4CDM");
|
|
418 |
String pubType = getValue(record, "PubType");
|
|
419 |
String place_of_publication = getValue(record, "place_of_publication");
|
|
420 |
String volume_and_page = getValue(record, "volume_and_page");
|
|
421 |
|
|
422 |
if (!CdmUtils.nullSafeEqual(name.getNameCache(), nameCache)){
|
|
423 |
logger.warn(row + "Unexpected nameCache: " + nameCache);
|
|
424 |
}
|
|
425 |
if (!CdmUtils.nullSafeEqual(name.getTitleCache(), fullName)){
|
|
426 |
logger.warn(row + "Unexpected titleCache: <->" + name.getTitleCache());
|
|
427 |
}
|
|
428 |
if (isBlank(genusHybrid) == name.isMonomHybrid()){
|
|
429 |
logger.warn(row + "Unexpected genus hybrid: " + genusHybrid);
|
|
430 |
}
|
|
431 |
if (!CdmUtils.nullSafeEqual(name.getGenusOrUninomial(),genus)){
|
|
432 |
logger.warn(row + "Unexpected genus: " + genus);
|
|
433 |
}if (isBlank(speciesHybrid) == name.isBinomHybrid()){
|
|
434 |
logger.warn(row + "Unexpected species hybrid: " + speciesHybrid);
|
|
435 |
}
|
|
436 |
if (!CdmUtils.nullSafeEqual(name.getSpecificEpithet(),species)){
|
|
437 |
logger.warn(row + "Unexpected species epithet: " + name.getSpecificEpithet() +"<->"+ species);
|
|
438 |
}
|
|
439 |
if (!CdmUtils.nullSafeEqual(name.getInfraSpecificEpithet(), infraspecies)){
|
|
440 |
logger.warn(row + "Unexpected infraspecific epithet: " + name.getInfraSpecificEpithet() +"<->"+ infraspecies);
|
|
441 |
}
|
|
442 |
if (!CdmUtils.nullSafeEqual(name.getAuthorshipCache(),authors)){
|
|
443 |
logger.warn(row + "Unexpected authors: " + name.getAuthorshipCache() +"<->"+ authors);
|
|
444 |
}
|
|
445 |
Rank rank = state.getTransformer().getRankByKey(rankStr);
|
|
446 |
if (!rank.equals(name.getRank())){
|
|
447 |
logger.warn(row + "Unexpected rank: " + rankStr);
|
|
448 |
}
|
|
449 |
|
|
450 |
Reference nomRef = name.getNomenclaturalReference();
|
|
451 |
if (nomRef == null){
|
|
452 |
if (fullCitation != null){
|
|
453 |
NonViralNameParserImpl parser = new NonViralNameParserImpl();
|
|
454 |
TaxonName parsedName = parser.parseReferencedName(fullCitation, NomenclaturalCode.ICNAFP, rank);
|
|
455 |
if (parsedName.getNomenclaturalReference() != null){
|
|
456 |
name.setNomenclaturalReference(parsedName.getNomenclaturalReference());
|
|
457 |
logger.warn(row + "Nom.ref. was missing. Taken from Kew");
|
|
458 |
}else{
|
|
459 |
logger.warn(row + "Nom. ref. is missing or can not be parsed");
|
|
460 |
}
|
|
461 |
}else{
|
|
462 |
logger.warn(row + "NomRef is missing.");
|
|
463 |
}
|
|
464 |
}else{
|
|
465 |
if ("A".equals(pubType) && nomRef.getType() != ReferenceType.Article){
|
|
466 |
logger.warn(row + "Unexpected nomref type: " + pubType + "<->" + nomRef.getType().toString());
|
|
467 |
}
|
|
468 |
if ("B".equals(pubType) && nomRef.getType() != ReferenceType.Book){
|
|
469 |
logger.warn(row + "Unexpected nomref type: " + pubType + "<->" + nomRef.getType().toString());
|
|
470 |
}
|
|
471 |
year = normalizeYear(year);
|
|
472 |
if (!CdmUtils.nullSafeEqual(year, nomRef.getDatePublishedString())){
|
|
473 |
logger.warn(row + "Unexpected year: " + year + "<->" + nomRef.getDatePublishedString());
|
|
474 |
}
|
|
475 |
if (!name.getFullTitleCache().contains(volume_and_page)){
|
|
476 |
logger.warn(row + "volume_and_page not found in fullTitleCache: " + name.getFullTitleCache() +"<->"+ volume_and_page);
|
|
477 |
}
|
|
478 |
if (!name.getFullTitleCache().contains(place_of_publication)){
|
|
479 |
logger.warn(row + "place_of_publication not found in fullTitleCache: " + name.getFullTitleCache() +"<->"+ place_of_publication);
|
|
480 |
}
|
|
481 |
}
|
|
482 |
if ("subsp.".equals(infraSpecRank) && !rank.equals(Rank.SUBSPECIES())){
|
|
483 |
logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank);
|
|
484 |
}else if ("var.".equals(infraSpecRank) && !rank.equals(Rank.VARIETY())){
|
|
485 |
logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank);
|
|
486 |
}else if ("f.".equals(infraSpecRank) && !rank.equals(Rank.FORM())){
|
|
487 |
logger.warn(row + "Unexpected infraspec marker: " + infraSpecRank);
|
|
488 |
}
|
|
489 |
|
|
490 |
}
|
|
491 |
|
|
492 |
private String normalizeYear(String year) {
|
|
493 |
if (year == null){
|
|
494 |
return null;
|
|
495 |
}else if (year.contains("\" [")){
|
|
496 |
String[] split = year.split("\" \\[");
|
|
497 |
year = split[1].replace("]","") + " [" + split[0]+"\"]";
|
|
498 |
}else if ("?".equals(year)){
|
|
499 |
return null;
|
|
500 |
}
|
|
501 |
return year;
|
|
502 |
}
|
|
503 |
|
|
504 |
@Override
|
|
505 |
protected void secondPass(SimpleExcelTaxonImportState<CaryoAizoaceaeExcelImportConfigurator> state) {
|
|
506 |
Map<String, String> record = state.getOriginalRecord();
|
|
507 |
int line = state.getCurrentLine();
|
|
508 |
String fullName = getValue(record, KEW_F_NAME4CDM_LINK);
|
|
509 |
String status = getValue(record, TAXON_STATUS);
|
|
510 |
String sourceId = getValue(record, PLANT_NAME_ID);
|
|
511 |
String accId = getValue(record, ACCEPTED_PLANT_NAME_ID);
|
|
512 |
String accName = getValue(record, "AcceptedName");
|
|
513 |
String basionymId = getValue(record, "basionym_plant_name_id");
|
|
514 |
String homotypicSynonym = getValue(record, "homotypic_synonym");
|
|
515 |
|
|
516 |
String row = String.valueOf(line) + "("+fullName+"): ";
|
|
517 |
try {
|
|
518 |
if ((line % 100) == 0){
|
|
519 |
newTransaction(state);
|
|
520 |
System.out.println(line);
|
|
521 |
}
|
|
522 |
|
|
523 |
if("Misapplied".equals(status)){
|
|
524 |
return;
|
|
525 |
}else if (neglectedRecords.contains(sourceId)){
|
|
526 |
logger.info(row + "Record ignored.");
|
|
527 |
return;
|
|
528 |
}
|
|
529 |
|
|
530 |
UUID uuid = taxonMapping.get(sourceId);
|
|
531 |
TaxonBase<?> taxonBase = getTaxonService().find(uuid);
|
|
532 |
if (taxonBase == null){
|
|
533 |
logger.warn(row + "taxonBase not found: " + sourceId);
|
|
534 |
return;
|
|
535 |
}
|
|
536 |
|
|
537 |
UUID accUuid = taxonMapping.get(accId);
|
|
538 |
boolean hasAccepted = !sourceId.equals(accId);
|
|
539 |
|
|
540 |
Taxon accTaxon = null;
|
|
541 |
TaxonNode parent = null;
|
|
542 |
Taxon child = null;
|
|
543 |
Synonym syn = null;
|
|
544 |
boolean isSynonymAccepted = false;
|
|
545 |
|
|
546 |
if(accId == null){
|
|
547 |
logger.info(row + "accID is null");
|
|
548 |
child = CdmBase.deproxy(taxonBase, Taxon.class);
|
|
549 |
}else if(hasAccepted){
|
|
550 |
TaxonBase<?> accTaxonBase = getTaxonService().find(accUuid);
|
|
551 |
if (accTaxonBase == null){
|
|
552 |
logger.warn(row + "acctaxon not found: " + accId + "; " + accName);
|
|
553 |
}else if(!accTaxonBase.isInstanceOf(Taxon.class)){
|
|
554 |
logger.warn(row + "acctaxon is synonym: " + accId + "; " + accName);
|
|
555 |
isSynonymAccepted = true;
|
|
556 |
}else{
|
|
557 |
accTaxon = CdmBase.deproxy(accTaxonBase, Taxon.class);
|
|
558 |
if (!accTaxon.getName().getTitleCache().equals(accName)){
|
|
559 |
logger.warn(row + "Accepted name differs: " + accName +" <-> "+ accTaxon.getName().getTitleCache());
|
|
560 |
}
|
|
561 |
}
|
|
562 |
}else if (sourceId.equals(accId)){
|
|
563 |
if (!taxonBase.isInstanceOf(Taxon.class)){
|
|
564 |
logger.warn(row + "child not of class Taxon: " + sourceId);
|
|
565 |
}else{
|
|
566 |
Rank rank = taxonBase.getName().getRank();
|
|
567 |
child = CdmBase.deproxy(taxonBase, Taxon.class);
|
|
568 |
if(rank.equals(Rank.GENUS())){
|
|
569 |
parent = getFamily();
|
|
570 |
}else if (rank.equals(Rank.SPECIES())){
|
|
571 |
String genus = child.getName().getGenusOrUninomial();
|
|
572 |
UUID parentUuid = taxonMapping.get(genus);
|
|
573 |
parent = getParent(parentUuid, row);
|
|
574 |
}else if (rank.isLower(Rank.SPECIES())){
|
|
575 |
String speciesName = child.getName().getGenusOrUninomial() + " " + child.getName().getSpecificEpithet();
|
|
576 |
UUID parentUuid = taxonMapping.get(speciesName);
|
|
577 |
parent = getParent(parentUuid, row);
|
|
578 |
}
|
|
579 |
}
|
|
580 |
}
|
|
581 |
|
|
582 |
if (taxonBase.isInstanceOf(Synonym.class)){
|
|
583 |
syn = CdmBase.deproxy(taxonBase, Synonym.class);
|
|
584 |
}
|
|
585 |
|
|
586 |
if ("Accepted".equals(status)){
|
|
587 |
if (parent == null){
|
|
588 |
logger.warn(row + "Parent is missing. Taxon is moved to 'unresolved' instead'");
|
|
589 |
parent = unresolvedParent();
|
|
590 |
}
|
|
591 |
if (child == null){
|
|
592 |
logger.warn(row + "Child is missing. Taxon not imported.");
|
|
593 |
}else{
|
|
594 |
if (!child.getTaxonNodes().isEmpty()){
|
|
595 |
if(!child.getName().getRank().equals(Rank.GENUS())){
|
|
596 |
logger.warn(row + "Taxon already has a parent. Taxon not attached to any further parent taxon.");
|
|
597 |
}
|
|
598 |
}else{
|
|
599 |
addChild(parent, child, row);
|
|
600 |
}
|
|
601 |
}
|
|
602 |
}else if ("Synonym".equals(status)){
|
|
603 |
if(accTaxon == null){
|
|
604 |
if(isSynonymAccepted){
|
|
605 |
logger.warn(row + "Synonym added to 'unresolved' as accepted taxon is synonym itself.");
|
|
606 |
}else if (accId != null){
|
|
607 |
logger.warn(row + "Accepted taxon for synonym unexpectedly does not exist (it seems not to be a synonym itself). Synonym moved to 'unresolved'");
|
|
608 |
}else{
|
|
609 |
logger.warn(row + "No accepted taxon given for synonym. Therefore taxon moved to 'unresolved'");
|
|
610 |
}
|
|
611 |
if(accId != null){
|
|
612 |
child = Taxon.NewInstance(syn.getName(), syn.getSec());
|
|
613 |
}
|
|
614 |
addChild(unresolvedParent(), child, row);
|
|
615 |
getTaxonService().deleteSynonym(syn, new SynonymDeletionConfigurator());
|
|
616 |
}else{
|
|
617 |
accTaxon.addSynonym(syn, SynonymType.SYNONYM_OF());
|
|
618 |
}
|
|
619 |
}else if ("Unplaced".equals(status)){
|
|
620 |
parent = unresolvedParent();
|
|
621 |
addChild(parent, child, row);
|
|
622 |
}else if ("Artificial Hybrid".equals(status)){
|
|
623 |
parent = hybridParent();
|
|
624 |
addChild(parent, child, row);
|
|
625 |
}else if ("Orthographic".equals(status)){
|
|
626 |
if(accTaxon == null){
|
|
627 |
logger.warn(row + "'Orthographic' taxon has no acc taxon");
|
|
628 |
}else{
|
|
629 |
accTaxon.addSynonym(syn, SynonymType.SYNONYM_OF());
|
|
630 |
}
|
|
631 |
}else if("Illegitimate".equals(status) || "Invalid".equals(status)){
|
|
632 |
if (hasAccepted){
|
|
633 |
if(accTaxon == null){
|
|
634 |
logger.warn(row + "accepted taxon for illegitimate or invalid taxon not found");
|
|
635 |
}else{
|
|
636 |
accTaxon.addSynonym(syn, SynonymType.SYNONYM_OF());
|
|
637 |
}
|
|
638 |
}else{
|
|
639 |
addChild(unresolvedParent(), child, row);
|
|
640 |
}
|
|
641 |
}else{
|
|
642 |
logger.warn(row + "Unhandled status: " + status);
|
|
643 |
}
|
|
644 |
|
|
645 |
if (basionymId != null){
|
|
646 |
UUID basionymUuid = taxonMapping.get(basionymId);
|
|
647 |
TaxonBase<?> basionymTaxon = getTaxonService().find(basionymUuid);
|
|
648 |
if (basionymTaxon != null){
|
|
649 |
if (hasSameAcceptedTaxon(taxonBase, basionymTaxon)){
|
|
650 |
if (taxonBase.getName().getBasionym() == null){
|
|
651 |
taxonBase.getName().addBasionym(basionymTaxon.getName());
|
|
652 |
}
|
|
653 |
}else{
|
|
654 |
logger.warn(row + "Basionym has not same accepted taxon and therefore was ignored.");
|
|
655 |
}
|
|
656 |
}else{
|
|
657 |
logger.warn(row + "Basionym "+basionymId+" not found.");
|
|
658 |
}
|
|
659 |
}
|
|
660 |
} catch (Exception e) {
|
|
661 |
logger.error(row + "Error.");
|
|
662 |
e.printStackTrace();
|
|
663 |
}
|
|
664 |
}
|
|
665 |
|
|
666 |
private boolean hasSameAcceptedTaxon(TaxonBase<?> taxonBase, TaxonBase<?> basionymTaxon) {
|
|
667 |
if (taxonBase.isInstanceOf(Synonym.class)){
|
|
668 |
taxonBase = CdmBase.deproxy(taxonBase, Synonym.class);
|
|
669 |
}
|
|
670 |
if (basionymTaxon.isInstanceOf(Synonym.class)){
|
|
671 |
basionymTaxon = CdmBase.deproxy(basionymTaxon, Synonym.class);
|
|
672 |
}
|
|
673 |
return taxonBase.equals(basionymTaxon);
|
|
674 |
}
|
|
675 |
|
|
676 |
private TaxonNode getParent(UUID parentUuid, String row) {
|
|
677 |
if(parentUuid == null){
|
|
678 |
logger.warn(row + "Parent uuid is null. No parent found.");
|
|
679 |
return null;
|
|
680 |
}
|
|
681 |
TaxonBase<?> pTaxon = getTaxonService().find(parentUuid);
|
|
682 |
if (pTaxon == null){
|
|
683 |
logger.warn(row + "No parent found for parent UUID. This should not happen.");
|
|
684 |
return null;
|
|
685 |
}
|
|
686 |
if (pTaxon.isInstanceOf(Synonym.class)){
|
|
687 |
logger.warn(row + "Parent is synonym");
|
|
688 |
return null;
|
|
689 |
}else{
|
|
690 |
Taxon ptax = CdmBase.deproxy(pTaxon, Taxon.class);
|
|
691 |
if(ptax.getTaxonNodes().isEmpty()){
|
|
692 |
logger.warn(row + "Parent has no node yet");
|
|
693 |
return null;
|
|
694 |
}else {
|
|
695 |
if(ptax.getTaxonNodes().size()>1){
|
|
696 |
logger.warn("Parent has >1 nodes. Take arbitrary one");
|
|
697 |
}
|
|
698 |
return ptax.getTaxonNodes().iterator().next();
|
|
699 |
}
|
|
700 |
}
|
|
701 |
}
|
|
702 |
|
|
703 |
private void addChild(TaxonNode parent, Taxon child, String row) {
|
|
704 |
if (parent == null){
|
|
705 |
logger.warn(row + "Parent is null");
|
|
706 |
}else if (child == null){
|
|
707 |
logger.warn(row + "Child is null");
|
|
708 |
}else{
|
|
709 |
if (!child.getTaxonNodes().isEmpty()){
|
|
710 |
TaxonNode childNode = child.getTaxonNodes().iterator().next();
|
|
711 |
if (childNode.getParent() != null && childNode.getParent().equals(parent)){
|
|
712 |
logger.info(row + "Parent-child relation exists already.");
|
|
713 |
}else{
|
|
714 |
logger.warn(row + "Child already has different parent. Parent-child relation not added.");
|
|
715 |
}
|
|
716 |
}else{
|
|
717 |
TaxonNode node = parent.addChildTaxon(child, null, null);
|
|
718 |
getTaxonNodeService().saveOrUpdate(node);
|
|
719 |
}
|
|
720 |
}
|
|
721 |
}
|
|
722 |
|
|
723 |
private TaxonNode getFamily(){
|
|
724 |
UUID uuid = UUID.fromString("0334809a-aa20-447d-add9-138194f80f56");
|
|
725 |
TaxonNode aizoaceae = getTaxonNodeService().find(uuid);
|
|
726 |
return aizoaceae;
|
|
727 |
}
|
|
728 |
|
|
729 |
private TaxonNode hybridParent(){
|
|
730 |
UUID uuid = UUID.fromString("2fae0fa1-758a-4fcb-bb6c-a2bd11f40641");
|
|
731 |
TaxonNode hybridParent = getTaxonNodeService().find(uuid);
|
|
732 |
return hybridParent;
|
|
733 |
}
|
|
734 |
private TaxonNode unresolvedParent(){
|
|
735 |
UUID uuid = UUID.fromString("accb1ff6-5748-4b18-b529-9368c331a38d");
|
|
736 |
TaxonNode unresolvedParent = getTaxonNodeService().find(uuid);
|
|
737 |
return unresolvedParent;
|
|
738 |
}
|
|
739 |
}
|
ref #8853 Aizoaceae import more or less ready