1 |
|
/**
|
2 |
|
* Copyright (C) 2007 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.taxonx;
|
10 |
|
|
11 |
|
import java.util.ArrayList;
|
12 |
|
import java.util.HashMap;
|
13 |
|
import java.util.List;
|
14 |
|
import java.util.Map;
|
15 |
|
import java.util.Set;
|
16 |
|
|
17 |
|
import org.apache.logging.log4j.LogManager;import org.apache.logging.log4j.Logger;
|
18 |
|
import org.jdom.Element;
|
19 |
|
import org.jdom.Namespace;
|
20 |
|
import org.springframework.stereotype.Component;
|
21 |
|
import org.springframework.transaction.TransactionStatus;
|
22 |
|
|
23 |
|
import eu.etaxonomy.cdm.api.service.ICommonService;
|
24 |
|
import eu.etaxonomy.cdm.api.service.INameService;
|
25 |
|
import eu.etaxonomy.cdm.api.service.ITaxonService;
|
26 |
|
import eu.etaxonomy.cdm.api.service.pager.Pager;
|
27 |
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
28 |
|
import eu.etaxonomy.cdm.io.common.CdmImportBase;
|
29 |
|
import eu.etaxonomy.cdm.io.common.IImportConfigurator;
|
30 |
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
31 |
|
import eu.etaxonomy.cdm.model.agent.Person;
|
32 |
|
import eu.etaxonomy.cdm.model.name.INonViralName;
|
33 |
|
import eu.etaxonomy.cdm.model.name.NameTypeDesignationStatus;
|
34 |
|
import eu.etaxonomy.cdm.model.name.Rank;
|
35 |
|
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus;
|
36 |
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
37 |
|
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
38 |
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
39 |
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
40 |
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
41 |
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
42 |
|
|
43 |
|
|
44 |
|
/**
|
45 |
|
* @author a.mueller
|
46 |
|
* @since 29.07.2008
|
47 |
|
*/
|
48 |
|
@Component
|
49 |
|
public class TaxonXNomenclatureImport
|
50 |
|
extends CdmImportBase<TaxonXImportConfigurator, TaxonXImportState> {
|
51 |
|
|
52 |
|
private static final long serialVersionUID = 796115831082828758L;
|
53 |
|
private static final Logger logger = LogManager.getLogger(TaxonXNomenclatureImport.class);
|
54 |
|
|
55 |
|
@SuppressWarnings("unused")
|
56 |
|
private static int modCount = 10000;
|
57 |
|
|
58 |
|
public TaxonXNomenclatureImport(){
|
59 |
|
super();
|
60 |
|
}
|
61 |
|
|
62 |
|
@Override
|
63 |
|
public boolean doCheck(TaxonXImportState state){
|
64 |
|
boolean result = true;
|
65 |
|
logger.warn("Checking for Types not yet implemented");
|
66 |
|
//result &= checkArticlesWithoutJournal(bmiConfig);
|
67 |
|
//result &= checkPartOfJournal(bmiConfig);
|
68 |
|
|
69 |
|
return result;
|
70 |
|
}
|
71 |
|
|
72 |
|
@Override
|
73 |
|
public void doInvoke(TaxonXImportState state){
|
74 |
|
logger.info("start make Nomenclature ...");
|
75 |
|
TransactionStatus tx = startTransaction();
|
76 |
|
TaxonXImportConfigurator config = state.getConfig();
|
77 |
|
Element root = config.getSourceRoot();
|
78 |
|
Namespace nsTaxonx = root.getNamespace();
|
79 |
|
|
80 |
|
//for testing only
|
81 |
|
Taxon taxon = getTaxon(config);
|
82 |
|
boolean isChanged = false;
|
83 |
|
|
84 |
|
Element elTaxonBody = root.getChild("taxonxBody", nsTaxonx);
|
85 |
|
Element elTreatment = elTaxonBody.getChild("treatment", nsTaxonx);
|
86 |
|
Element elNomenclature = elTreatment.getChild("nomenclature", nsTaxonx);
|
87 |
|
|
88 |
|
//isChanged |= doCollectionEvent(txConfig, elNomenclature, nsTaxonx, taxon);
|
89 |
|
|
90 |
|
if (taxon != null && taxon.getName() != null && elNomenclature != null){
|
91 |
|
isChanged |= doNomenclaturalType(config, elNomenclature, nsTaxonx, taxon.getName());
|
92 |
|
List<Element> elSynonymyList = new ArrayList<Element>();
|
93 |
|
elSynonymyList.addAll(elNomenclature.getChildren("synonomy", nsTaxonx));
|
94 |
|
elSynonymyList.addAll(elNomenclature.getChildren("synonymy", nsTaxonx)); //wrong spelling in TaxonX-Schema
|
95 |
|
for (Element elSynonymy : elSynonymyList){
|
96 |
|
String synonymName = elSynonymy.getChildTextTrim("name");
|
97 |
|
if (elSynonymy.getChild("type", nsTaxonx) != null || elSynonymy.getChild("type_loc", nsTaxonx) != null){
|
98 |
|
Synonym synonym = getSynonym(config, taxon, synonymName);
|
99 |
|
if (synonym != null){
|
100 |
|
isChanged |= doNomenclaturalType(config, elSynonymy, nsTaxonx, synonym.getName());
|
101 |
|
}
|
102 |
|
}
|
103 |
|
}
|
104 |
|
}
|
105 |
|
|
106 |
|
|
107 |
|
if (isChanged){
|
108 |
|
getTaxonService().save(taxon);
|
109 |
|
}
|
110 |
|
commitTransaction(tx);
|
111 |
|
return;
|
112 |
|
}
|
113 |
|
|
114 |
|
private Synonym getSynonym(TaxonXImportConfigurator config, Taxon taxon, String synName){
|
115 |
|
Synonym result = null;
|
116 |
|
unlazySynonym(config, taxon);
|
117 |
|
Set<Synonym> synList = taxon.getSynonyms();
|
118 |
|
for (Synonym syn : synList){
|
119 |
|
TaxonName nameBase = syn.getName();
|
120 |
|
if (nameBase != null){
|
121 |
|
if (nameBase.isNonViral()){
|
122 |
|
if (nameBase.getNameCache().equals(synName)){
|
123 |
|
return syn; //only first synonym is returned
|
124 |
|
}
|
125 |
|
}
|
126 |
|
}
|
127 |
|
}
|
128 |
|
logger.warn("Synonym ("+synName+ ")not found for taxon " + taxon.getTitleCache() + getBracketSourceName(config));
|
129 |
|
return result;
|
130 |
|
}
|
131 |
|
|
132 |
|
private Taxon getTaxon(TaxonXImportConfigurator config){
|
133 |
|
Taxon result;
|
134 |
|
// result = Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(null), null);
|
135 |
|
//ICommonService commonService =config.getCdmAppController().getCommonService();
|
136 |
|
ICommonService commonService = getCommonService();
|
137 |
|
String originalSourceId = config.getOriginalSourceId();
|
138 |
|
String namespace = config.getOriginalSourceTaxonNamespace();
|
139 |
|
result = commonService.getSourcedObjectByIdInSource(Taxon.class, originalSourceId , namespace);
|
140 |
|
if (result == null){
|
141 |
|
logger.warn("Taxon (id: " + originalSourceId + ", namespace: " + namespace + ") could not be found");
|
142 |
|
}
|
143 |
|
return result;
|
144 |
|
}
|
145 |
|
|
146 |
|
/* (non-Javadoc)
|
147 |
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
148 |
|
*/
|
149 |
|
@Override
|
150 |
|
protected boolean isIgnore(TaxonXImportState state){
|
151 |
|
return ! state.getConfig().isDoTypes();
|
152 |
|
}
|
153 |
|
|
154 |
|
/**
|
155 |
|
*
|
156 |
|
* Reads the collection_event tag, creates the according data and stores it.
|
157 |
|
* TODO under work
|
158 |
|
* @param elNomenclature
|
159 |
|
* @param nsTaxonx
|
160 |
|
* @param taxonBase
|
161 |
|
* @return
|
162 |
|
*/
|
163 |
|
private boolean doNomenclaturalType(TaxonXImportConfigurator config, Element elNomenclature, Namespace nsTaxonx, TaxonName taxonName){
|
164 |
|
boolean success = true;
|
165 |
|
if (taxonName == null){
|
166 |
|
logger.warn("taxonName is null");
|
167 |
|
return false;
|
168 |
|
}
|
169 |
|
if (elNomenclature == null){
|
170 |
|
logger.warn("elNomenclature is null");
|
171 |
|
return false;
|
172 |
|
}
|
173 |
|
|
174 |
|
|
175 |
|
Element elType = elNomenclature.getChild("type", nsTaxonx);
|
176 |
|
Element elTypeLoc = elNomenclature.getChild("type_loc", nsTaxonx);
|
177 |
|
|
178 |
|
if (elType != null || elTypeLoc != null){
|
179 |
|
unlazyTypeDesignation(config, taxonName);
|
180 |
|
|
181 |
|
if (taxonName.isInfraGeneric() || taxonName.isSupraGeneric() || taxonName.isGenus()){
|
182 |
|
success &= doNameType(elType, taxonName, config);
|
183 |
|
}else{
|
184 |
|
success &= doSpecimenType(config, elType, elTypeLoc, taxonName);
|
185 |
|
|
186 |
|
|
187 |
|
}
|
188 |
|
return success;
|
189 |
|
}
|
190 |
|
return false;
|
191 |
|
}
|
192 |
|
|
193 |
|
|
194 |
|
private boolean doSpecimenType(TaxonXImportConfigurator config, Element elType, Element elTypeLoc, TaxonName taxonName){
|
195 |
|
Reference citation = null;
|
196 |
|
String citationMicroReference = null;
|
197 |
|
String originalNameString = null;
|
198 |
|
boolean isNotDesignated = true;
|
199 |
|
boolean addToAllHomotypicNames = true;
|
200 |
|
|
201 |
|
SimpleSpecimen simpleSpecimen = SimpleSpecimen.NewInstance();
|
202 |
|
//elType
|
203 |
|
if (elType != null){
|
204 |
|
doElType(elType, simpleSpecimen, config);
|
205 |
|
}//elType
|
206 |
|
|
207 |
|
//typeLoc
|
208 |
|
HashMap<DerivedUnit, SpecimenTypeDesignationStatus> typeLocMap = null;
|
209 |
|
if (elTypeLoc != null){
|
210 |
|
typeLocMap = doElTypeLoc(elTypeLoc, simpleSpecimen, taxonName, config);
|
211 |
|
}
|
212 |
|
if (typeLocMap != null && typeLocMap.size() >0){
|
213 |
|
for (DerivedUnit specimen : typeLocMap.keySet()){
|
214 |
|
SpecimenTypeDesignationStatus status = typeLocMap.get(specimen);
|
215 |
|
taxonName.addSpecimenTypeDesignation(specimen, status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames);
|
216 |
|
}
|
217 |
|
}else{ // no type_loc
|
218 |
|
SpecimenTypeDesignationStatus status = null;
|
219 |
|
taxonName.addSpecimenTypeDesignation(simpleSpecimen.getSpecimen(), status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames);
|
220 |
|
}
|
221 |
|
return true;
|
222 |
|
}
|
223 |
|
|
224 |
|
private boolean doElType(Element elType, SimpleSpecimen simpleSpecimen, TaxonXImportConfigurator config){
|
225 |
|
//type
|
226 |
|
String text = elType.getTextNormalize();
|
227 |
|
if (text.endsWith(";")){
|
228 |
|
text = text + " ";
|
229 |
|
}
|
230 |
|
String[] type = text.split(";");
|
231 |
|
if (type.length != 3 ){
|
232 |
|
if (text.equals("")){
|
233 |
|
logger.info("<nomenclature><type> is empty: " + getBracketSourceName(config));
|
234 |
|
}else{
|
235 |
|
logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
236 |
|
}
|
237 |
|
simpleSpecimen.setTitleCache(elType.getTextNormalize());
|
238 |
|
}else{
|
239 |
|
String strLocality = type[0].trim();
|
240 |
|
if (! "".equals(strLocality)){
|
241 |
|
// simpleSpecimen.setLocality(strLocality);
|
242 |
|
}
|
243 |
|
|
244 |
|
String strCollector = type[1].trim();
|
245 |
|
if (! "".equals(strCollector)){
|
246 |
|
AgentBase collector = Person.NewTitledInstance(strCollector);
|
247 |
|
// simpleSpecimen.setCollector(collector);
|
248 |
|
}
|
249 |
|
|
250 |
|
String strCollectorNumber = type[2].trim();
|
251 |
|
if (! "".equals(strCollectorNumber)){
|
252 |
|
// simpleSpecimen.setCollectorsNumber(strCollectorNumber);
|
253 |
|
}
|
254 |
|
|
255 |
|
String title = CdmUtils.concat(" ", new String[]{strLocality, strCollector, strCollectorNumber});
|
256 |
|
simpleSpecimen.setTitleCache(title);
|
257 |
|
}
|
258 |
|
return true;
|
259 |
|
}
|
260 |
|
|
261 |
|
private boolean doNameType(Element elType, TaxonName taxonName, TaxonXImportConfigurator config){
|
262 |
|
boolean success = true;
|
263 |
|
//type
|
264 |
|
String text = elType.getTextNormalize();
|
265 |
|
logger.info("Type: " + text);
|
266 |
|
if (text.endsWith(";")){
|
267 |
|
text = text + " ";
|
268 |
|
}
|
269 |
|
String[] type = text.split(";");
|
270 |
|
if (type.length != 3 ){
|
271 |
|
if (text.equals("")){
|
272 |
|
logger.info("<nomenclature><type> is empty: " + getBracketSourceName(config));
|
273 |
|
}else{
|
274 |
|
logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
275 |
|
}
|
276 |
|
success = false;
|
277 |
|
}else{
|
278 |
|
String statusStr = type[0].trim();
|
279 |
|
String taxonNameStr = type[1].trim();
|
280 |
|
String authorStr = type[2].trim();
|
281 |
|
NameTypeDesignationStatus status = getNameTypeStatus(statusStr);
|
282 |
|
/*boolean isLectoType = getIsLectoType(statusStr);*/
|
283 |
|
|
284 |
|
// if (status == null){
|
285 |
|
// logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
286 |
|
// success = false;
|
287 |
|
// }else{
|
288 |
|
// TaxonName childType = getChildrenNameType(taxonName, taxonNameStr, authorStr);
|
289 |
|
// if (childType != null){
|
290 |
|
// return doNameTypeDesignation(taxonName, childType, status);
|
291 |
|
// }else{
|
292 |
|
String[] epis = taxonNameStr.split(" ");
|
293 |
|
String uninomial = epis[0].trim();
|
294 |
|
String specEpi = epis[1].trim();
|
295 |
|
|
296 |
|
Pager<TaxonName> nameTypes = getNameService().searchNames(uninomial, null, specEpi, null, Rank.SPECIES(), null, null, null, null);
|
297 |
|
|
298 |
|
List<INonViralName> result = new ArrayList<>();
|
299 |
|
for (TaxonName nt : nameTypes.getRecords()){
|
300 |
|
if (compareAuthorship(nt, authorStr)){
|
301 |
|
result.add(nt);
|
302 |
|
success &= doNameTypeDesignation(taxonName, nt, status/*, isLectoType*/);
|
303 |
|
}else{
|
304 |
|
//TODO ?
|
305 |
|
}
|
306 |
|
}
|
307 |
|
if (result.size() > 1){
|
308 |
|
logger.warn("More than 1 name matches: " + text);
|
309 |
|
success = false;
|
310 |
|
}else if (result.size() == 0){
|
311 |
|
logger.warn("No name matches: " + text + "(" + config.getSourceNameString() + ")");
|
312 |
|
success = false;
|
313 |
|
}
|
314 |
|
// }
|
315 |
|
// }
|
316 |
|
}
|
317 |
|
return success;
|
318 |
|
}
|
319 |
|
|
320 |
|
|
321 |
|
// private TaxonName getChildrenNameType(TaxonName name, String typeStr, String authorStr){
|
322 |
|
// TaxonName result = null;
|
323 |
|
// Set<TaxonBase> list = name.getTaxonBases();
|
324 |
|
// for (TaxonBase taxonBase : list){
|
325 |
|
// Taxon taxon;
|
326 |
|
// if (taxonBase.isInstanceOf(Taxon.class)){
|
327 |
|
// taxon = CdmBase.deproxy(taxonBase, Taxon.class);
|
328 |
|
// }else{
|
329 |
|
// Synonym syn = CdmBase.deproxy(taxonBase, Synonym.class);
|
330 |
|
// taxon = syn.getAcceptedTaxa().iterator().next();
|
331 |
|
// }
|
332 |
|
// Set<Taxon> children = taxon.getTaxonomicChildren();
|
333 |
|
// for (Taxon child: children){
|
334 |
|
// INonViralName childName = child.getName();
|
335 |
|
// if (childName.getNameCache().equals(typeStr)){
|
336 |
|
// if (compareAuthorship(childName, authorStr)){
|
337 |
|
// return childName;
|
338 |
|
// }
|
339 |
|
// }
|
340 |
|
// }
|
341 |
|
// }
|
342 |
|
// return result;
|
343 |
|
// }
|
344 |
|
|
345 |
|
private boolean compareAuthorship(INonViralName typeName, String authorStr){
|
346 |
|
boolean result = false;
|
347 |
|
authorStr = authorStr.replaceAll("\\s+and\\s+", "&");
|
348 |
|
authorStr = authorStr.replaceAll("\\s*", "");
|
349 |
|
authorStr = authorStr.replaceAll("\\.$", "");
|
350 |
|
String typeCache = typeName.getAuthorshipCache().replaceAll("\\s*", "");
|
351 |
|
typeCache = typeCache.replaceAll("\\.$", "");
|
352 |
|
if (authorStr.equals(typeCache)){
|
353 |
|
return true;
|
354 |
|
}else{
|
355 |
|
logger.info(" Authors different: " + authorStr + " <-> " + typeCache);
|
356 |
|
}
|
357 |
|
return result;
|
358 |
|
}
|
359 |
|
|
360 |
|
private NameTypeDesignationStatus getNameTypeStatus(String statusString){
|
361 |
|
//FIXME some types (not further defined types) do not exist yet
|
362 |
|
if (true){
|
363 |
|
return null;
|
364 |
|
}
|
365 |
|
if (statusString.trim().equalsIgnoreCase("Type")){
|
366 |
|
return NameTypeDesignationStatus.ORIGINAL_DESIGNATION();
|
367 |
|
}else if (statusString.trim().equalsIgnoreCase("Lectotype")){
|
368 |
|
return NameTypeDesignationStatus.LECTOTYPE();
|
369 |
|
}else if (statusString.trim().equalsIgnoreCase("Holotype")){
|
370 |
|
logger.warn("Holotype does not yet exist in CDM");
|
371 |
|
return NameTypeDesignationStatus.NOT_APPLICABLE();
|
372 |
|
}else if (statusString.trim().equalsIgnoreCase("paratype")){
|
373 |
|
logger.warn("paratype does not yet exist in CDM");
|
374 |
|
return NameTypeDesignationStatus.NOT_APPLICABLE();
|
375 |
|
}
|
376 |
|
else{
|
377 |
|
logger.warn("Status not recognized: " + statusString);
|
378 |
|
return null;
|
379 |
|
}
|
380 |
|
}
|
381 |
|
|
382 |
|
private boolean getIsLectoType(String statusString){
|
383 |
|
//FIXME may be deleted once getNameTypeStatus works finde
|
384 |
|
if (statusString.trim().equals("Lectotype")){
|
385 |
|
return true;
|
386 |
|
}else{
|
387 |
|
return false;
|
388 |
|
}
|
389 |
|
}
|
390 |
|
|
391 |
|
|
392 |
|
private boolean doNameTypeDesignation(TaxonName name, TaxonName type, NameTypeDesignationStatus status/*, boolean isLectoType*/){
|
393 |
|
Reference citation = null;
|
394 |
|
String citationMicroReference = null;
|
395 |
|
String originalNameString = null;
|
396 |
|
boolean addToAllHomotypicNames = true;
|
397 |
|
|
398 |
|
// name.addNameTypeDesignation(type, citation, citationMicroReference, originalNameString, status, addToAllHomotypicNames);
|
399 |
|
name.addNameTypeDesignation(type, citation, citationMicroReference, originalNameString,status, false, false, /*isLectoType, */false, addToAllHomotypicNames);
|
400 |
|
return true;
|
401 |
|
}
|
402 |
|
|
403 |
|
/**
|
404 |
|
* Reads the typeLoc element split in parts for eacht type (holo, iso,...)
|
405 |
|
* @param elTypeLoc
|
406 |
|
* @param simpleSpecimen
|
407 |
|
* @param taxonName
|
408 |
|
* @param config
|
409 |
|
* @return
|
410 |
|
*/
|
411 |
|
private HashMap<DerivedUnit, SpecimenTypeDesignationStatus> doElTypeLoc(Element elTypeLoc,
|
412 |
|
SimpleSpecimen simpleSpecimen,
|
413 |
|
TaxonName taxonName,
|
414 |
|
TaxonXImportConfigurator config){
|
415 |
|
|
416 |
|
HashMap<DerivedUnit, SpecimenTypeDesignationStatus> result = new HashMap<DerivedUnit, SpecimenTypeDesignationStatus>();
|
417 |
|
|
418 |
|
String typeLocFullString = elTypeLoc.getTextTrim();
|
419 |
|
typeLocFullString = typeLocFullString.replace("(", "").replace(")", "");
|
420 |
|
String[] typeLocStatusList = typeLocFullString.split(";");
|
421 |
|
|
422 |
|
DerivedUnit originalSpecimen = simpleSpecimen.getSpecimen();
|
423 |
|
|
424 |
|
|
425 |
|
for (String typeLocStatus : typeLocStatusList){
|
426 |
|
typeLocStatus = typeLocStatus.trim();
|
427 |
|
int pos = typeLocStatus.indexOf(" ");
|
428 |
|
if (pos == -1){
|
429 |
|
logger.warn("Unknown format or empty type_loc : '" +typeLocStatus + "'" + getBracketSourceName(config));
|
430 |
|
result.put(originalSpecimen, null);
|
431 |
|
}else{
|
432 |
|
String statusString = typeLocStatus.substring(0,pos);
|
433 |
|
SpecimenTypeDesignationStatus status = getStatusByStatusString(statusString.trim(), config);
|
434 |
|
//TODO
|
435 |
|
//String[] collectionStrings = typeLocStatus.substring(pos).split(",");
|
436 |
|
String tmpCollString = typeLocStatus.substring(pos).trim();
|
437 |
|
//for(String collectionString : collectionStrings){
|
438 |
|
if (tmpCollString.contains("typ")){
|
439 |
|
logger.warn("Is this really only a collection string? : " + tmpCollString + getBracketSourceName(config));
|
440 |
|
}
|
441 |
|
DerivedUnit specimen;
|
442 |
|
specimen = (DerivedUnit)originalSpecimen.clone();
|
443 |
|
String title = originalSpecimen.getTitleCache();
|
444 |
|
title = title + "(" + tmpCollString + ")";
|
445 |
|
specimen.setTitleCache(title, true );
|
446 |
|
result.put(specimen, status);
|
447 |
|
//}
|
448 |
|
}
|
449 |
|
}
|
450 |
|
|
451 |
|
return result;
|
452 |
|
}
|
453 |
|
|
454 |
|
/**
|
455 |
|
*
|
456 |
|
* Reads the collection_event tag, creates the according data and stores it.
|
457 |
|
* TODO under work
|
458 |
|
* @param elNomenclature
|
459 |
|
* @param nsTaxonx
|
460 |
|
* @param taxonBase
|
461 |
|
* @return
|
462 |
|
*/
|
463 |
|
private boolean doCollectionEvent(TaxonXImportConfigurator config, Element elNomenclature, Namespace nsTaxonx, TaxonBase taxonBase){
|
464 |
|
boolean result = false;
|
465 |
|
if (elNomenclature == null){
|
466 |
|
return false;
|
467 |
|
}
|
468 |
|
Element elCollectionEvent = elNomenclature.getChild("collection_event", nsTaxonx);
|
469 |
|
if (elCollectionEvent == null){
|
470 |
|
return result;
|
471 |
|
}
|
472 |
|
Element elLocality = elCollectionEvent.getChild("locality", nsTaxonx);
|
473 |
|
Element elType = elCollectionEvent.getChild("type", nsTaxonx);
|
474 |
|
Element elTypeLoc = elCollectionEvent.getChild("type_loc", nsTaxonx);
|
475 |
|
|
476 |
|
//locality
|
477 |
|
SimpleSpecimen simpleSpecimen = SimpleSpecimen.NewInstance();
|
478 |
|
String locality = elLocality.getTextNormalize();
|
479 |
|
if (! "".equals(locality)){
|
480 |
|
simpleSpecimen.setLocality(locality);
|
481 |
|
}
|
482 |
|
|
483 |
|
//type
|
484 |
|
String[] type = elType.getTextNormalize().split(" ");
|
485 |
|
if (type.length != 2 ){
|
486 |
|
logger.warn("<collecion_even><type> is of unsupported format: " + elType.getTextNormalize());
|
487 |
|
}else{
|
488 |
|
AgentBase collector = Person.NewTitledInstance(type[0]);
|
489 |
|
simpleSpecimen.setCollector(collector);
|
490 |
|
|
491 |
|
String collectorNumber = type[1];
|
492 |
|
simpleSpecimen.setCollectorsNumber(collectorNumber);
|
493 |
|
}
|
494 |
|
|
495 |
|
//typeLoc
|
496 |
|
String typeLocFullString = elTypeLoc.getTextTrim();
|
497 |
|
typeLocFullString = typeLocFullString.replace("(", "").replace(")", "");
|
498 |
|
String[] typeLocStatusList = typeLocFullString.split(";");
|
499 |
|
|
500 |
|
DerivedUnit originalSpecimen = simpleSpecimen.getSpecimen();
|
501 |
|
|
502 |
|
//TODO special character ?, �, !
|
503 |
|
|
504 |
|
for (String typeLocStatus : typeLocStatusList){
|
505 |
|
typeLocStatus = typeLocStatus.trim();
|
506 |
|
int pos = typeLocStatus.indexOf(" ");
|
507 |
|
if (pos == -1){
|
508 |
|
logger.warn("Unknown format: " + typeLocStatus);
|
509 |
|
}else{
|
510 |
|
String statusString = typeLocStatus.substring(0,pos);
|
511 |
|
SpecimenTypeDesignationStatus status = getStatusByStatusString(statusString.trim(), config);
|
512 |
|
String[] collectionStrings = typeLocStatus.substring(pos).split(",");
|
513 |
|
for(String collectionString : collectionStrings){
|
514 |
|
if (taxonBase != null){
|
515 |
|
TaxonName taxonName = taxonBase.getName();
|
516 |
|
if (taxonName != null){
|
517 |
|
Reference citation = null;
|
518 |
|
String citationMicroReference = null;
|
519 |
|
String originalNameString = null;
|
520 |
|
boolean isNotDesignated = true;
|
521 |
|
boolean addToAllHomotypicNames = true;
|
522 |
|
DerivedUnit specimen = (DerivedUnit)originalSpecimen.clone();
|
523 |
|
unlazyTypeDesignation(config, taxonName);
|
524 |
|
taxonName.addSpecimenTypeDesignation(specimen, status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames);
|
525 |
|
result = true;
|
526 |
|
}
|
527 |
|
}
|
528 |
|
}
|
529 |
|
}
|
530 |
|
}
|
531 |
|
return result;
|
532 |
|
}
|
533 |
|
|
534 |
|
|
535 |
|
private static Map<String, SpecimenTypeDesignationStatus> statusMap;
|
536 |
|
private static void fillTypeStatusMap(){
|
537 |
|
statusMap = new HashMap<String, SpecimenTypeDesignationStatus>();
|
538 |
|
statusMap.put("epitype", SpecimenTypeDesignationStatus.EPITYPE());
|
539 |
|
statusMap.put("holotype", SpecimenTypeDesignationStatus.HOLOTYPE());
|
540 |
|
statusMap.put("iconotype", SpecimenTypeDesignationStatus.ICONOTYPE());
|
541 |
|
statusMap.put("isotype", SpecimenTypeDesignationStatus.ISOTYPE());
|
542 |
|
statusMap.put("isoneotype", SpecimenTypeDesignationStatus.ISONEOTYPE());
|
543 |
|
statusMap.put("isosyntype", SpecimenTypeDesignationStatus.ISOSYNTYPE());
|
544 |
|
statusMap.put("isolectotype", SpecimenTypeDesignationStatus.ISOLECTOTYPE());
|
545 |
|
statusMap.put("lectotype", SpecimenTypeDesignationStatus.LECTOTYPE());
|
546 |
|
statusMap.put("syntype", SpecimenTypeDesignationStatus.SYNTYPE());
|
547 |
|
statusMap.put("paratype", SpecimenTypeDesignationStatus.PARATYPE());
|
548 |
|
statusMap.put("neotype", SpecimenTypeDesignationStatus.NEOTYPE());
|
549 |
|
statusMap.put("isoepitype", SpecimenTypeDesignationStatus.ISOEPITYPE());
|
550 |
|
statusMap.put("originalmaterial", SpecimenTypeDesignationStatus.ORIGINAL_MATERIAL());
|
551 |
|
statusMap.put("paralectotype", SpecimenTypeDesignationStatus.PARALECTOTYPE());
|
552 |
|
statusMap.put("paraneotype", SpecimenTypeDesignationStatus.PARANEOTYPE());
|
553 |
|
statusMap.put("phototype", SpecimenTypeDesignationStatus.PHOTOTYPE());
|
554 |
|
statusMap.put("secondsteplectotype", SpecimenTypeDesignationStatus.SECOND_STEP_LECTOTYPE());
|
555 |
|
statusMap.put("secondstepneotype", SpecimenTypeDesignationStatus.SECOND_STEP_NEOTYPE());
|
556 |
|
statusMap.put("type", null);
|
557 |
|
}
|
558 |
|
|
559 |
|
|
560 |
|
//TODO move to TypeDesignation class
|
561 |
|
/**
|
562 |
|
* Returns the typeDesignationStatus according to a type designation status string
|
563 |
|
* @param statusString
|
564 |
|
* @return TypeDesignationStatus
|
565 |
|
*/
|
566 |
|
private static SpecimenTypeDesignationStatus getStatusByStatusString(String statusString, TaxonXImportConfigurator config){
|
567 |
|
SpecimenTypeDesignationStatus result = null;
|
568 |
|
if (statusString == null || "".equals(statusString.trim())){
|
569 |
|
return null;
|
570 |
|
}
|
571 |
|
statusString = statusString.trim().toLowerCase();
|
572 |
|
statusString = statusString.replace("typi", "typus");
|
573 |
|
statusString = statusString.replace("typus", "type");
|
574 |
|
statusString = statusString.replace("types", "type");
|
575 |
|
statusString = statusString.toLowerCase();
|
576 |
|
|
577 |
|
if (statusMap == null){
|
578 |
|
fillTypeStatusMap();
|
579 |
|
}
|
580 |
|
result = statusMap.get(statusString);
|
581 |
|
if (statusString.equals("type")){
|
582 |
|
logger.info("No type designation type" + getBracketSourceName(config));
|
583 |
|
}else if (result == null){
|
584 |
|
logger.warn("Unknown type status string: " + statusString + getBracketSourceName(config));
|
585 |
|
}
|
586 |
|
return result;
|
587 |
|
}
|
588 |
|
|
589 |
|
|
590 |
|
/**
|
591 |
|
* TODO Preliminary to avoid laizy loading errors
|
592 |
|
*/
|
593 |
|
private void unlazyTypeDesignation(TaxonXImportConfigurator config, TaxonName taxonName){
|
594 |
|
TransactionStatus txStatus = startTransaction();
|
595 |
|
//INameService taxonNameService = config.getCdmAppController().getNameService();
|
596 |
|
INameService taxonNameService = getNameService();
|
597 |
|
|
598 |
|
taxonNameService.save(taxonName);
|
599 |
|
Set<TaxonName> typifiedNames = taxonName.getHomotypicalGroup().getTypifiedNames();
|
600 |
|
for(TaxonName typifiedName: typifiedNames){
|
601 |
|
typifiedName.getTypeDesignations().size();
|
602 |
|
}
|
603 |
|
//taxonNameService.saveTaxonName(taxonName);
|
604 |
|
commitTransaction(txStatus);
|
605 |
|
}
|
606 |
|
|
607 |
|
/**
|
608 |
|
* TODO Preliminary to avoid laizy loading errors
|
609 |
|
*/
|
610 |
|
private void unlazySynonym(IImportConfigurator config, Taxon taxon){
|
611 |
|
TransactionStatus txStatus = startTransaction();
|
612 |
|
ITaxonService taxonService = getTaxonService();
|
613 |
|
taxonService.save(taxon);
|
614 |
|
Set<Synonym> synonyms = taxon.getSynonyms();
|
615 |
|
logger.debug(synonyms.size());
|
616 |
|
//taxonService.saveTaxon(taxon);
|
617 |
|
commitTransaction(txStatus);
|
618 |
|
}
|
619 |
|
|
620 |
|
private static String getBracketSourceName(TaxonXImportConfigurator config){
|
621 |
|
return "(" + config.getSourceNameString() + ")";
|
622 |
|
}
|
623 |
|
|
624 |
|
|
625 |
|
}
|
|
1 |
/**
|
|
2 |
* Copyright (C) 2007 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.taxonx;
|
|
10 |
|
|
11 |
import java.util.ArrayList;
|
|
12 |
import java.util.HashMap;
|
|
13 |
import java.util.List;
|
|
14 |
import java.util.Map;
|
|
15 |
import java.util.Set;
|
|
16 |
|
|
17 |
import org.apache.logging.log4j.LogManager;import org.apache.logging.log4j.Logger;
|
|
18 |
import org.jdom.Element;
|
|
19 |
import org.jdom.Namespace;
|
|
20 |
import org.springframework.stereotype.Component;
|
|
21 |
import org.springframework.transaction.TransactionStatus;
|
|
22 |
|
|
23 |
import eu.etaxonomy.cdm.api.service.ICommonService;
|
|
24 |
import eu.etaxonomy.cdm.api.service.INameService;
|
|
25 |
import eu.etaxonomy.cdm.api.service.ITaxonService;
|
|
26 |
import eu.etaxonomy.cdm.api.service.pager.Pager;
|
|
27 |
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
28 |
import eu.etaxonomy.cdm.io.common.CdmImportBase;
|
|
29 |
import eu.etaxonomy.cdm.io.common.IImportConfigurator;
|
|
30 |
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
|
31 |
import eu.etaxonomy.cdm.model.agent.Person;
|
|
32 |
import eu.etaxonomy.cdm.model.name.INonViralName;
|
|
33 |
import eu.etaxonomy.cdm.model.name.NameTypeDesignationStatus;
|
|
34 |
import eu.etaxonomy.cdm.model.name.Rank;
|
|
35 |
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus;
|
|
36 |
import eu.etaxonomy.cdm.model.name.TaxonName;
|
|
37 |
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
|
38 |
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
39 |
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
|
40 |
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
|
41 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
42 |
|
|
43 |
|
|
44 |
/**
|
|
45 |
* @author a.mueller
|
|
46 |
* @since 29.07.2008
|
|
47 |
*/
|
|
48 |
@Component
|
|
49 |
public class TaxonXNomenclatureImport
|
|
50 |
extends CdmImportBase<TaxonXImportConfigurator, TaxonXImportState> {
|
|
51 |
|
|
52 |
private static final long serialVersionUID = 796115831082828758L;
|
|
53 |
private static final Logger logger = LogManager.getLogger(TaxonXNomenclatureImport.class);
|
|
54 |
|
|
55 |
@SuppressWarnings("unused")
|
|
56 |
private static int modCount = 10000;
|
|
57 |
|
|
58 |
public TaxonXNomenclatureImport(){
|
|
59 |
super();
|
|
60 |
}
|
|
61 |
|
|
62 |
@Override
|
|
63 |
public boolean doCheck(TaxonXImportState state){
|
|
64 |
boolean result = true;
|
|
65 |
logger.warn("Checking for Types not yet implemented");
|
|
66 |
//result &= checkArticlesWithoutJournal(bmiConfig);
|
|
67 |
//result &= checkPartOfJournal(bmiConfig);
|
|
68 |
|
|
69 |
return result;
|
|
70 |
}
|
|
71 |
|
|
72 |
@Override
|
|
73 |
public void doInvoke(TaxonXImportState state){
|
|
74 |
logger.info("start make Nomenclature ...");
|
|
75 |
TransactionStatus tx = startTransaction();
|
|
76 |
TaxonXImportConfigurator config = state.getConfig();
|
|
77 |
Element root = config.getSourceRoot();
|
|
78 |
Namespace nsTaxonx = root.getNamespace();
|
|
79 |
|
|
80 |
//for testing only
|
|
81 |
Taxon taxon = getTaxon(config);
|
|
82 |
boolean isChanged = false;
|
|
83 |
|
|
84 |
Element elTaxonBody = root.getChild("taxonxBody", nsTaxonx);
|
|
85 |
Element elTreatment = elTaxonBody.getChild("treatment", nsTaxonx);
|
|
86 |
Element elNomenclature = elTreatment.getChild("nomenclature", nsTaxonx);
|
|
87 |
|
|
88 |
//isChanged |= doCollectionEvent(txConfig, elNomenclature, nsTaxonx, taxon);
|
|
89 |
|
|
90 |
if (taxon != null && taxon.getName() != null && elNomenclature != null){
|
|
91 |
isChanged |= doNomenclaturalType(config, elNomenclature, nsTaxonx, taxon.getName());
|
|
92 |
List<Element> elSynonymyList = new ArrayList<Element>();
|
|
93 |
elSynonymyList.addAll(elNomenclature.getChildren("synonomy", nsTaxonx));
|
|
94 |
elSynonymyList.addAll(elNomenclature.getChildren("synonymy", nsTaxonx)); //wrong spelling in TaxonX-Schema
|
|
95 |
for (Element elSynonymy : elSynonymyList){
|
|
96 |
String synonymName = elSynonymy.getChildTextTrim("name");
|
|
97 |
if (elSynonymy.getChild("type", nsTaxonx) != null || elSynonymy.getChild("type_loc", nsTaxonx) != null){
|
|
98 |
Synonym synonym = getSynonym(config, taxon, synonymName);
|
|
99 |
if (synonym != null){
|
|
100 |
isChanged |= doNomenclaturalType(config, elSynonymy, nsTaxonx, synonym.getName());
|
|
101 |
}
|
|
102 |
}
|
|
103 |
}
|
|
104 |
}
|
|
105 |
|
|
106 |
|
|
107 |
if (isChanged){
|
|
108 |
getTaxonService().save(taxon);
|
|
109 |
}
|
|
110 |
commitTransaction(tx);
|
|
111 |
return;
|
|
112 |
}
|
|
113 |
|
|
114 |
private Synonym getSynonym(TaxonXImportConfigurator config, Taxon taxon, String synName){
|
|
115 |
Synonym result = null;
|
|
116 |
unlazySynonym(config, taxon);
|
|
117 |
Set<Synonym> synList = taxon.getSynonyms();
|
|
118 |
for (Synonym syn : synList){
|
|
119 |
TaxonName nameBase = syn.getName();
|
|
120 |
if (nameBase != null){
|
|
121 |
if (nameBase.isNonViral()){
|
|
122 |
if (nameBase.getNameCache().equals(synName)){
|
|
123 |
return syn; //only first synonym is returned
|
|
124 |
}
|
|
125 |
}
|
|
126 |
}
|
|
127 |
}
|
|
128 |
logger.warn("Synonym ("+synName+ ")not found for taxon " + taxon.getTitleCache() + getBracketSourceName(config));
|
|
129 |
return result;
|
|
130 |
}
|
|
131 |
|
|
132 |
private Taxon getTaxon(TaxonXImportConfigurator config){
|
|
133 |
Taxon result;
|
|
134 |
// result = Taxon.NewInstance(TaxonNameFactory.NewBotanicalInstance(null), null);
|
|
135 |
//ICommonService commonService =config.getCdmAppController().getCommonService();
|
|
136 |
ICommonService commonService = getCommonService();
|
|
137 |
String originalSourceId = config.getOriginalSourceId();
|
|
138 |
String namespace = config.getOriginalSourceTaxonNamespace();
|
|
139 |
result = commonService.getSourcedObjectByIdInSource(Taxon.class, originalSourceId , namespace);
|
|
140 |
if (result == null){
|
|
141 |
logger.warn("Taxon (id: " + originalSourceId + ", namespace: " + namespace + ") could not be found");
|
|
142 |
}
|
|
143 |
return result;
|
|
144 |
}
|
|
145 |
|
|
146 |
/* (non-Javadoc)
|
|
147 |
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
|
148 |
*/
|
|
149 |
@Override
|
|
150 |
protected boolean isIgnore(TaxonXImportState state){
|
|
151 |
return ! state.getConfig().isDoTypes();
|
|
152 |
}
|
|
153 |
|
|
154 |
/**
|
|
155 |
*
|
|
156 |
* Reads the collection_event tag, creates the according data and stores it.
|
|
157 |
* TODO under work
|
|
158 |
* @param elNomenclature
|
|
159 |
* @param nsTaxonx
|
|
160 |
* @param taxonBase
|
|
161 |
* @return
|
|
162 |
*/
|
|
163 |
private boolean doNomenclaturalType(TaxonXImportConfigurator config, Element elNomenclature, Namespace nsTaxonx, TaxonName taxonName){
|
|
164 |
boolean success = true;
|
|
165 |
if (taxonName == null){
|
|
166 |
logger.warn("taxonName is null");
|
|
167 |
return false;
|
|
168 |
}
|
|
169 |
if (elNomenclature == null){
|
|
170 |
logger.warn("elNomenclature is null");
|
|
171 |
return false;
|
|
172 |
}
|
|
173 |
|
|
174 |
|
|
175 |
Element elType = elNomenclature.getChild("type", nsTaxonx);
|
|
176 |
Element elTypeLoc = elNomenclature.getChild("type_loc", nsTaxonx);
|
|
177 |
|
|
178 |
if (elType != null || elTypeLoc != null){
|
|
179 |
unlazyTypeDesignation(config, taxonName);
|
|
180 |
|
|
181 |
if (taxonName.isInfraGeneric() || taxonName.isSupraGeneric() || taxonName.isGenus()){
|
|
182 |
success &= doNameType(elType, taxonName, config);
|
|
183 |
}else{
|
|
184 |
success &= doSpecimenType(config, elType, elTypeLoc, taxonName);
|
|
185 |
|
|
186 |
|
|
187 |
}
|
|
188 |
return success;
|
|
189 |
}
|
|
190 |
return false;
|
|
191 |
}
|
|
192 |
|
|
193 |
|
|
194 |
private boolean doSpecimenType(TaxonXImportConfigurator config, Element elType, Element elTypeLoc, TaxonName taxonName){
|
|
195 |
Reference citation = null;
|
|
196 |
String citationMicroReference = null;
|
|
197 |
String originalNameString = null;
|
|
198 |
boolean isNotDesignated = true;
|
|
199 |
boolean addToAllHomotypicNames = true;
|
|
200 |
|
|
201 |
SimpleSpecimen simpleSpecimen = SimpleSpecimen.NewInstance();
|
|
202 |
//elType
|
|
203 |
if (elType != null){
|
|
204 |
doElType(elType, simpleSpecimen, config);
|
|
205 |
}//elType
|
|
206 |
|
|
207 |
//typeLoc
|
|
208 |
HashMap<DerivedUnit, SpecimenTypeDesignationStatus> typeLocMap = null;
|
|
209 |
if (elTypeLoc != null){
|
|
210 |
typeLocMap = doElTypeLoc(elTypeLoc, simpleSpecimen, taxonName, config);
|
|
211 |
}
|
|
212 |
if (typeLocMap != null && typeLocMap.size() >0){
|
|
213 |
for (DerivedUnit specimen : typeLocMap.keySet()){
|
|
214 |
SpecimenTypeDesignationStatus status = typeLocMap.get(specimen);
|
|
215 |
taxonName.addSpecimenTypeDesignation(specimen, status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames);
|
|
216 |
}
|
|
217 |
}else{ // no type_loc
|
|
218 |
SpecimenTypeDesignationStatus status = null;
|
|
219 |
taxonName.addSpecimenTypeDesignation(simpleSpecimen.getSpecimen(), status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames);
|
|
220 |
}
|
|
221 |
return true;
|
|
222 |
}
|
|
223 |
|
|
224 |
private boolean doElType(Element elType, SimpleSpecimen simpleSpecimen, TaxonXImportConfigurator config){
|
|
225 |
//type
|
|
226 |
String text = elType.getTextNormalize();
|
|
227 |
if (text.endsWith(";")){
|
|
228 |
text = text + " ";
|
|
229 |
}
|
|
230 |
String[] type = text.split(";");
|
|
231 |
if (type.length != 3 ){
|
|
232 |
if (text.equals("")){
|
|
233 |
logger.info("<nomenclature><type> is empty: " + getBracketSourceName(config));
|
|
234 |
}else{
|
|
235 |
logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
|
236 |
}
|
|
237 |
simpleSpecimen.setTitleCache(elType.getTextNormalize());
|
|
238 |
}else{
|
|
239 |
String strLocality = type[0].trim();
|
|
240 |
if (! "".equals(strLocality)){
|
|
241 |
// simpleSpecimen.setLocality(strLocality);
|
|
242 |
}
|
|
243 |
|
|
244 |
String strCollector = type[1].trim();
|
|
245 |
if (! "".equals(strCollector)){
|
|
246 |
AgentBase collector = Person.NewTitledInstance(strCollector);
|
|
247 |
// simpleSpecimen.setCollector(collector);
|
|
248 |
}
|
|
249 |
|
|
250 |
String strCollectorNumber = type[2].trim();
|
|
251 |
if (! "".equals(strCollectorNumber)){
|
|
252 |
// simpleSpecimen.setCollectorsNumber(strCollectorNumber);
|
|
253 |
}
|
|
254 |
|
|
255 |
String title = CdmUtils.concat(" ", new String[]{strLocality, strCollector, strCollectorNumber});
|
|
256 |
simpleSpecimen.setTitleCache(title);
|
|
257 |
}
|
|
258 |
return true;
|
|
259 |
}
|
|
260 |
|
|
261 |
private boolean doNameType(Element elType, TaxonName taxonName, TaxonXImportConfigurator config){
|
|
262 |
boolean success = true;
|
|
263 |
//type
|
|
264 |
String text = elType.getTextNormalize();
|
|
265 |
logger.info("Type: " + text);
|
|
266 |
if (text.endsWith(";")){
|
|
267 |
text = text + " ";
|
|
268 |
}
|
|
269 |
String[] type = text.split(";");
|
|
270 |
if (type.length != 3 ){
|
|
271 |
if (text.equals("")){
|
|
272 |
logger.info("<nomenclature><type> is empty: " + getBracketSourceName(config));
|
|
273 |
}else{
|
|
274 |
logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
|
275 |
}
|
|
276 |
success = false;
|
|
277 |
}else{
|
|
278 |
String statusStr = type[0].trim();
|
|
279 |
String taxonNameStr = type[1].trim();
|
|
280 |
String authorStr = type[2].trim();
|
|
281 |
NameTypeDesignationStatus status = getNameTypeStatus(statusStr);
|
|
282 |
/*boolean isLectoType = getIsLectoType(statusStr);*/
|
|
283 |
|
|
284 |
// if (status == null){
|
|
285 |
// logger.warn("<nomenclature><type> is of unsupported format: " + elType.getTextNormalize() + getBracketSourceName(config));
|
|
286 |
// success = false;
|
|
287 |
// }else{
|
|
288 |
// TaxonName childType = getChildrenNameType(taxonName, taxonNameStr, authorStr);
|
|
289 |
// if (childType != null){
|
|
290 |
// return doNameTypeDesignation(taxonName, childType, status);
|
|
291 |
// }else{
|
|
292 |
String[] epis = taxonNameStr.split(" ");
|
|
293 |
String uninomial = epis[0].trim();
|
|
294 |
String specEpi = epis[1].trim();
|
|
295 |
|
|
296 |
Pager<TaxonName> nameTypes = getNameService().searchNames(uninomial, null, specEpi, null, Rank.SPECIES(), null, null, null, null);
|
|
297 |
|
|
298 |
List<INonViralName> result = new ArrayList<>();
|
|
299 |
for (TaxonName nt : nameTypes.getRecords()){
|
|
300 |
if (compareAuthorship(nt, authorStr)){
|
|
301 |
result.add(nt);
|
|
302 |
success &= doNameTypeDesignation(taxonName, nt, status/*, isLectoType*/);
|
|
303 |
}else{
|
|
304 |
//TODO ?
|
|
305 |
}
|
|
306 |
}
|
|
307 |
if (result.size() > 1){
|
|
308 |
logger.warn("More than 1 name matches: " + text);
|
|
309 |
success = false;
|
|
310 |
}else if (result.size() == 0){
|
|
311 |
logger.warn("No name matches: " + text + "(" + config.getSourceNameString() + ")");
|
|
312 |
success = false;
|
|
313 |
}
|
|
314 |
// }
|
|
315 |
// }
|
|
316 |
}
|
|
317 |
return success;
|
|
318 |
}
|
|
319 |
|
|
320 |
|
|
321 |
// private TaxonName getChildrenNameType(TaxonName name, String typeStr, String authorStr){
|
|
322 |
// TaxonName result = null;
|
|
323 |
// Set<TaxonBase> list = name.getTaxonBases();
|
|
324 |
// for (TaxonBase taxonBase : list){
|
|
325 |
// Taxon taxon;
|
|
326 |
// if (taxonBase.isInstanceOf(Taxon.class)){
|
|
327 |
// taxon = CdmBase.deproxy(taxonBase, Taxon.class);
|
|
328 |
// }else{
|
|
329 |
// Synonym syn = CdmBase.deproxy(taxonBase, Synonym.class);
|
|
330 |
// taxon = syn.getAcceptedTaxa().iterator().next();
|
|
331 |
// }
|
|
332 |
// Set<Taxon> children = taxon.getTaxonomicChildren();
|
|
333 |
// for (Taxon child: children){
|
|
334 |
// INonViralName childName = child.getName();
|
|
335 |
// if (childName.getNameCache().equals(typeStr)){
|
|
336 |
// if (compareAuthorship(childName, authorStr)){
|
|
337 |
// return childName;
|
|
338 |
// }
|
|
339 |
// }
|
|
340 |
// }
|
|
341 |
// }
|
|
342 |
// return result;
|
|
343 |
// }
|
|
344 |
|
|
345 |
private boolean compareAuthorship(INonViralName typeName, String authorStr){
|
|
346 |
boolean result = false;
|
|
347 |
authorStr = authorStr.replaceAll("\\s+and\\s+", "&");
|
|
348 |
authorStr = authorStr.replaceAll("\\s*", "");
|
|
349 |
authorStr = authorStr.replaceAll("\\.$", "");
|
|
350 |
String typeCache = typeName.getAuthorshipCache().replaceAll("\\s*", "");
|
|
351 |
typeCache = typeCache.replaceAll("\\.$", "");
|
|
352 |
if (authorStr.equals(typeCache)){
|
|
353 |
return true;
|
|
354 |
}else{
|
|
355 |
logger.info(" Authors different: " + authorStr + " <-> " + typeCache);
|
|
356 |
}
|
|
357 |
return result;
|
|
358 |
}
|
|
359 |
|
|
360 |
private NameTypeDesignationStatus getNameTypeStatus(String statusString){
|
|
361 |
//FIXME some types (not further defined types) do not exist yet
|
|
362 |
if (true){
|
|
363 |
return null;
|
|
364 |
}
|
|
365 |
if (statusString.trim().equalsIgnoreCase("Type")){
|
|
366 |
return NameTypeDesignationStatus.ORIGINAL_DESIGNATION();
|
|
367 |
}else if (statusString.trim().equalsIgnoreCase("Lectotype")){
|
|
368 |
return NameTypeDesignationStatus.LECTOTYPE();
|
|
369 |
}else if (statusString.trim().equalsIgnoreCase("Holotype")){
|
|
370 |
logger.warn("Holotype does not yet exist in CDM");
|
|
371 |
return NameTypeDesignationStatus.NOT_APPLICABLE();
|
|
372 |
}else if (statusString.trim().equalsIgnoreCase("paratype")){
|
|
373 |
logger.warn("paratype does not yet exist in CDM");
|
|
374 |
return NameTypeDesignationStatus.NOT_APPLICABLE();
|
|
375 |
}
|
|
376 |
else{
|
|
377 |
logger.warn("Status not recognized: " + statusString);
|
|
378 |
return null;
|
|
379 |
}
|
|
380 |
}
|
|
381 |
|
|
382 |
private boolean getIsLectoType(String statusString){
|
|
383 |
//FIXME may be deleted once getNameTypeStatus works finde
|
|
384 |
if (statusString.trim().equals("Lectotype")){
|
|
385 |
return true;
|
|
386 |
}else{
|
|
387 |
return false;
|
|
388 |
}
|
|
389 |
}
|
|
390 |
|
|
391 |
|
|
392 |
private boolean doNameTypeDesignation(TaxonName name, TaxonName type, NameTypeDesignationStatus status/*, boolean isLectoType*/){
|
|
393 |
Reference citation = null;
|
|
394 |
String citationMicroReference = null;
|
|
395 |
String originalNameString = null;
|
|
396 |
boolean addToAllHomotypicNames = true;
|
|
397 |
|
|
398 |
// name.addNameTypeDesignation(type, citation, citationMicroReference, originalNameString, status, addToAllHomotypicNames);
|
|
399 |
name.addNameTypeDesignation(type, citation, citationMicroReference, originalNameString,status, false, false, /*isLectoType, */false, addToAllHomotypicNames);
|
|
400 |
return true;
|
|
401 |
}
|
|
402 |
|
|
403 |
/**
|
|
404 |
* Reads the typeLoc element split in parts for eacht type (holo, iso,...)
|
|
405 |
* @param elTypeLoc
|
|
406 |
* @param simpleSpecimen
|
|
407 |
* @param taxonName
|
|
408 |
* @param config
|
|
409 |
* @return
|
|
410 |
*/
|
|
411 |
private HashMap<DerivedUnit, SpecimenTypeDesignationStatus> doElTypeLoc(Element elTypeLoc,
|
|
412 |
SimpleSpecimen simpleSpecimen,
|
|
413 |
TaxonName taxonName,
|
|
414 |
TaxonXImportConfigurator config){
|
|
415 |
|
|
416 |
HashMap<DerivedUnit, SpecimenTypeDesignationStatus> result = new HashMap<DerivedUnit, SpecimenTypeDesignationStatus>();
|
|
417 |
|
|
418 |
String typeLocFullString = elTypeLoc.getTextTrim();
|
|
419 |
typeLocFullString = typeLocFullString.replace("(", "").replace(")", "");
|
|
420 |
String[] typeLocStatusList = typeLocFullString.split(";");
|
|
421 |
|
|
422 |
DerivedUnit originalSpecimen = simpleSpecimen.getSpecimen();
|
|
423 |
|
|
424 |
|
|
425 |
for (String typeLocStatus : typeLocStatusList){
|
|
426 |
typeLocStatus = typeLocStatus.trim();
|
|
427 |
int pos = typeLocStatus.indexOf(" ");
|
|
428 |
if (pos == -1){
|
|
429 |
logger.warn("Unknown format or empty type_loc : '" +typeLocStatus + "'" + getBracketSourceName(config));
|
|
430 |
result.put(originalSpecimen, null);
|
|
431 |
}else{
|
|
432 |
String statusString = typeLocStatus.substring(0,pos);
|
|
433 |
SpecimenTypeDesignationStatus status = getStatusByStatusString(statusString.trim(), config);
|
|
434 |
//TODO
|
|
435 |
//String[] collectionStrings = typeLocStatus.substring(pos).split(",");
|
|
436 |
String tmpCollString = typeLocStatus.substring(pos).trim();
|
|
437 |
//for(String collectionString : collectionStrings){
|
|
438 |
if (tmpCollString.contains("typ")){
|
|
439 |
logger.warn("Is this really only a collection string? : " + tmpCollString + getBracketSourceName(config));
|
|
440 |
}
|
|
441 |
DerivedUnit specimen;
|
|
442 |
specimen = (DerivedUnit)originalSpecimen.clone();
|
|
443 |
String title = originalSpecimen.getTitleCache();
|
|
444 |
title = title + "(" + tmpCollString + ")";
|
|
445 |
specimen.setTitleCache(title, true );
|
|
446 |
result.put(specimen, status);
|
|
447 |
//}
|
|
448 |
}
|
|
449 |
}
|
|
450 |
|
|
451 |
return result;
|
|
452 |
}
|
|
453 |
|
|
454 |
/**
|
|
455 |
*
|
|
456 |
* Reads the collection_event tag, creates the according data and stores it.
|
|
457 |
* TODO under work
|
|
458 |
* @param elNomenclature
|
|
459 |
* @param nsTaxonx
|
|
460 |
* @param taxonBase
|
|
461 |
* @return
|
|
462 |
*/
|
|
463 |
private boolean doCollectionEvent(TaxonXImportConfigurator config, Element elNomenclature, Namespace nsTaxonx, TaxonBase taxonBase){
|
|
464 |
boolean result = false;
|
|
465 |
if (elNomenclature == null){
|
|
466 |
return false;
|
|
467 |
}
|
|
468 |
Element elCollectionEvent = elNomenclature.getChild("collection_event", nsTaxonx);
|
|
469 |
if (elCollectionEvent == null){
|
|
470 |
return result;
|
|
471 |
}
|
|
472 |
Element elLocality = elCollectionEvent.getChild("locality", nsTaxonx);
|
|
473 |
Element elType = elCollectionEvent.getChild("type", nsTaxonx);
|
|
474 |
Element elTypeLoc = elCollectionEvent.getChild("type_loc", nsTaxonx);
|
|
475 |
|
|
476 |
//locality
|
|
477 |
SimpleSpecimen simpleSpecimen = SimpleSpecimen.NewInstance();
|
|
478 |
String locality = elLocality.getTextNormalize();
|
|
479 |
if (! "".equals(locality)){
|
|
480 |
simpleSpecimen.setLocality(locality);
|
|
481 |
}
|
|
482 |
|
|
483 |
//type
|
|
484 |
String[] type = elType.getTextNormalize().split(" ");
|
|
485 |
if (type.length != 2 ){
|
|
486 |
logger.warn("<collecion_even><type> is of unsupported format: " + elType.getTextNormalize());
|
|
487 |
}else{
|
|
488 |
AgentBase collector = Person.NewTitledInstance(type[0]);
|
|
489 |
simpleSpecimen.setCollector(collector);
|
|
490 |
|
|
491 |
String collectorNumber = type[1];
|
|
492 |
simpleSpecimen.setCollectorsNumber(collectorNumber);
|
|
493 |
}
|
|
494 |
|
|
495 |
//typeLoc
|
cleanup