|
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
|
|
|
10
|
package eu.etaxonomy.cdm.io.redlist.gefaesspflanzen;
|
|
11
|
|
|
12
|
import java.sql.ResultSet;
|
|
13
|
import java.sql.SQLException;
|
|
14
|
import java.util.HashMap;
|
|
15
|
import java.util.HashSet;
|
|
16
|
import java.util.Map;
|
|
17
|
import java.util.Set;
|
|
18
|
import java.util.regex.Matcher;
|
|
19
|
import java.util.regex.Pattern;
|
|
20
|
|
|
21
|
import org.apache.log4j.Logger;
|
|
22
|
import org.springframework.stereotype.Component;
|
|
23
|
|
|
24
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
25
|
import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper;
|
|
26
|
import eu.etaxonomy.cdm.io.common.DbImportBase;
|
|
27
|
import eu.etaxonomy.cdm.io.common.IPartitionedIO;
|
|
28
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
|
29
|
import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
|
|
30
|
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException;
|
|
31
|
import eu.etaxonomy.cdm.model.agent.AgentBase;
|
|
32
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
|
33
|
import eu.etaxonomy.cdm.model.common.Annotation;
|
|
34
|
import eu.etaxonomy.cdm.model.common.AnnotationType;
|
|
35
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
36
|
import eu.etaxonomy.cdm.model.common.ExtensionType;
|
|
37
|
import eu.etaxonomy.cdm.model.common.Language;
|
|
38
|
import eu.etaxonomy.cdm.model.common.OrderedTermVocabulary;
|
|
39
|
import eu.etaxonomy.cdm.model.description.CommonTaxonName;
|
|
40
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
|
41
|
import eu.etaxonomy.cdm.model.name.ICultivarPlantName;
|
|
42
|
import eu.etaxonomy.cdm.model.name.INonViralName;
|
|
43
|
import eu.etaxonomy.cdm.model.name.ITaxonNameBase;
|
|
44
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
|
45
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus;
|
|
46
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
|
47
|
import eu.etaxonomy.cdm.model.name.Rank;
|
|
48
|
import eu.etaxonomy.cdm.model.name.RankClass;
|
|
49
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
|
50
|
import eu.etaxonomy.cdm.model.name.TaxonNameFactory;
|
|
51
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
|
52
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
|
53
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
54
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
|
55
|
|
|
56
|
/**
|
|
57
|
*
|
|
58
|
* @author pplitzner
|
|
59
|
* @since Mar 1, 2016
|
|
60
|
*
|
|
61
|
*/
|
|
62
|
|
|
63
|
@Component
|
|
64
|
@SuppressWarnings("serial")
|
|
65
|
public class RedListGefaesspflanzenImportNames extends DbImportBase<RedListGefaesspflanzenImportState, RedListGefaesspflanzenImportConfigurator> {
|
|
66
|
|
|
67
|
private static final Logger logger = Logger.getLogger(RedListGefaesspflanzenImportNames.class);
|
|
68
|
|
|
69
|
private static final String tableName = "Rote Liste Gefäßpflanzen";
|
|
70
|
|
|
71
|
private static final String pluralString = "names";
|
|
72
|
|
|
73
|
private static final boolean STRICT_TITLE_CHECK = false;
|
|
74
|
|
|
75
|
private ExtensionType extensionTypeFlor;
|
|
76
|
|
|
77
|
private ExtensionType extensionTypeAtlasIdx;
|
|
78
|
|
|
79
|
private ExtensionType extensionTypeKart;
|
|
80
|
|
|
81
|
private ExtensionType extensionTypeRl2015;
|
|
82
|
|
|
83
|
private ExtensionType extensionTypeEhrd;
|
|
84
|
|
|
85
|
private ExtensionType extensionTypeWissk;
|
|
86
|
|
|
87
|
public RedListGefaesspflanzenImportNames() {
|
|
88
|
super(tableName, pluralString);
|
|
89
|
}
|
|
90
|
|
|
91
|
@Override
|
|
92
|
protected String getIdQuery(RedListGefaesspflanzenImportState state) {
|
|
93
|
return "SELECT SEQNUM "
|
|
94
|
+ "FROM V_TAXATLAS_D20_EXPORT t "
|
|
95
|
+ " ORDER BY SEQNUM";
|
|
96
|
}
|
|
97
|
|
|
98
|
@Override
|
|
99
|
protected String getRecordQuery(RedListGefaesspflanzenImportConfigurator config) {
|
|
100
|
String result = " SELECT * "
|
|
101
|
+ " FROM V_TAXATLAS_D20_EXPORT t "
|
|
102
|
+ " WHERE t.SEQNUM IN (@IDSET)";
|
|
103
|
result = result.replace("@IDSET", IPartitionedIO.ID_LIST_TOKEN);
|
|
104
|
return result;
|
|
105
|
}
|
|
106
|
|
|
107
|
@Override
|
|
108
|
protected void doInvoke(RedListGefaesspflanzenImportState state) {
|
|
109
|
makeExtensionTypes();
|
|
110
|
super.doInvoke(state);
|
|
111
|
}
|
|
112
|
|
|
113
|
|
|
114
|
private void makeExtensionTypes() {
|
|
115
|
extensionTypeFlor = ExtensionType.NewInstance(RedListUtil.FLOR, RedListUtil.FLOR, "");
|
|
116
|
extensionTypeAtlasIdx = ExtensionType.NewInstance(RedListUtil.ATLAS_IDX, RedListUtil.ATLAS_IDX, "");
|
|
117
|
extensionTypeKart = ExtensionType.NewInstance(RedListUtil.KART, RedListUtil.KART, "");
|
|
118
|
extensionTypeRl2015 = ExtensionType.NewInstance(RedListUtil.RL2015, RedListUtil.RL2015, "");
|
|
119
|
extensionTypeEhrd = ExtensionType.NewInstance(RedListUtil.EHRD, RedListUtil.EHRD, "");
|
|
120
|
extensionTypeWissk = ExtensionType.NewInstance(RedListUtil.WISSK, RedListUtil.WISSK, "");
|
|
121
|
getTermService().saveOrUpdate(extensionTypeFlor);
|
|
122
|
getTermService().saveOrUpdate(extensionTypeAtlasIdx);
|
|
123
|
getTermService().saveOrUpdate(extensionTypeKart);
|
|
124
|
getTermService().saveOrUpdate(extensionTypeRl2015);
|
|
125
|
getTermService().saveOrUpdate(extensionTypeEhrd);
|
|
126
|
getTermService().saveOrUpdate(extensionTypeWissk);
|
|
127
|
}
|
|
128
|
|
|
129
|
@Override
|
|
130
|
public boolean doPartition(ResultSetPartitioner partitioner, RedListGefaesspflanzenImportState state) {
|
|
131
|
ResultSet rs = partitioner.getResultSet();
|
|
132
|
Set<ITaxonNameBase> namesToSave = new HashSet<>();
|
|
133
|
Set<TaxonBase> taxaToSave = new HashSet<>();
|
|
134
|
try {
|
|
135
|
while (rs.next()){
|
|
136
|
makeSingleNameAndTaxon(state, rs, namesToSave, taxaToSave);
|
|
137
|
|
|
138
|
}
|
|
139
|
} catch (SQLException e) {
|
|
140
|
e.printStackTrace();
|
|
141
|
}
|
|
142
|
|
|
143
|
getNameService().saveOrUpdate(TaxonName.castAndDeproxy(namesToSave));
|
|
144
|
getTaxonService().saveOrUpdate(taxaToSave);
|
|
145
|
return true;
|
|
146
|
}
|
|
147
|
|
|
148
|
private void makeSingleNameAndTaxon(RedListGefaesspflanzenImportState state, ResultSet rs, Set<ITaxonNameBase> namesToSave, Set<TaxonBase> taxaToSave)
|
|
149
|
throws SQLException {
|
|
150
|
long id = rs.getLong(RedListUtil.NAMNR);
|
|
151
|
String relationE = rs.getString(RedListUtil.E);
|
|
152
|
String relationW = rs.getString(RedListUtil.W);
|
|
153
|
String relationK = rs.getString(RedListUtil.K);
|
|
154
|
String relationAW = rs.getString(RedListUtil.AW);
|
|
155
|
String relationAO = rs.getString(RedListUtil.AO);
|
|
156
|
String relationR = rs.getString(RedListUtil.R);
|
|
157
|
String relationO = rs.getString(RedListUtil.O);
|
|
158
|
String relationS = rs.getString(RedListUtil.S);
|
|
159
|
|
|
160
|
//---NAME---
|
|
161
|
INonViralName name = importName(state, rs, namesToSave);
|
|
162
|
|
|
163
|
|
|
164
|
//--- AUTHORS ---
|
|
165
|
importAuthors(state, rs, name);
|
|
166
|
|
|
167
|
//---TAXON---
|
|
168
|
TaxonBase<?> taxonBase = importTaxon(rs, name, state);
|
|
169
|
if(taxonBase==null){
|
|
170
|
RedListUtil.logMessage(id, "!SERIOUS ERROR! Taxon for name "+name+" could not be created!", logger);
|
|
171
|
return;
|
|
172
|
}
|
|
173
|
|
|
174
|
//---CONCEPT RELATIONSHIPS---
|
|
175
|
//E, W, K, AW, AO, R, O, S
|
|
176
|
cloneTaxon(taxonBase, relationE, RedListUtil.CLASSIFICATION_NAMESPACE_E, taxaToSave, id, state);
|
|
177
|
cloneTaxon(taxonBase, relationW, RedListUtil.CLASSIFICATION_NAMESPACE_W, taxaToSave, id, state);
|
|
178
|
cloneTaxon(taxonBase, relationK, RedListUtil.CLASSIFICATION_NAMESPACE_K, taxaToSave, id, state);
|
|
179
|
cloneTaxon(taxonBase, relationAW, RedListUtil.CLASSIFICATION_NAMESPACE_AW, taxaToSave, id, state);
|
|
180
|
cloneTaxon(taxonBase, relationAO, RedListUtil.CLASSIFICATION_NAMESPACE_AO, taxaToSave, id, state);
|
|
181
|
cloneTaxon(taxonBase, relationR, RedListUtil.CLASSIFICATION_NAMESPACE_R, taxaToSave, id, state);
|
|
182
|
cloneTaxon(taxonBase, relationO, RedListUtil.CLASSIFICATION_NAMESPACE_O, taxaToSave, id, state);
|
|
183
|
cloneTaxon(taxonBase, relationS, RedListUtil.CLASSIFICATION_NAMESPACE_S, taxaToSave, id, state);
|
|
184
|
|
|
185
|
//NOTE: the source has to be added after cloning or otherwise the clone would also get the source
|
|
186
|
ImportHelper.setOriginalSource(taxonBase, state.getTransactionalSourceReference(), id, RedListUtil.TAXON_GESAMTLISTE_NAMESPACE);
|
|
187
|
taxaToSave.add(taxonBase);
|
|
188
|
}
|
|
189
|
|
|
190
|
private void cloneTaxon(final TaxonBase<?> gesamtListeTaxon, String relationString, String sourceNameSpace, Set<TaxonBase> taxaToSave, long id, RedListGefaesspflanzenImportState state){
|
|
191
|
if(CdmUtils.isNotBlank(relationString) && !relationString.equals(".")){
|
|
192
|
Taxon clonedTaxon = null;
|
|
193
|
|
|
194
|
if(gesamtListeTaxon.isInstanceOf(Taxon.class)){
|
|
195
|
clonedTaxon = HibernateProxyHelper.deproxy(gesamtListeTaxon.clone(), Taxon.class);
|
|
196
|
}
|
|
197
|
else if(gesamtListeTaxon.isInstanceOf(Synonym.class)){
|
|
198
|
clonedTaxon = Taxon.NewInstance(gesamtListeTaxon.getName(), gesamtListeTaxon.getSec());
|
|
199
|
}
|
|
200
|
else{
|
|
201
|
RedListUtil.logMessage(id, "Taxon base "+gesamtListeTaxon+" is neither taxon nor synonym! Taxon could not be cloned", logger);
|
|
202
|
return;
|
|
203
|
}
|
|
204
|
ImportHelper.setOriginalSource(clonedTaxon, state.getTransactionalSourceReference(), id, sourceNameSpace);
|
|
205
|
taxaToSave.add(clonedTaxon);
|
|
206
|
}
|
|
207
|
}
|
|
208
|
|
|
209
|
private TaxonBase<?> importTaxon(ResultSet rs, INonViralName name, RedListGefaesspflanzenImportState state) throws SQLException {
|
|
210
|
|
|
211
|
long id = rs.getLong(RedListUtil.NAMNR);
|
|
212
|
String taxNameString = rs.getString(RedListUtil.TAXNAME);
|
|
213
|
String epi1String = rs.getString(RedListUtil.EPI1);
|
|
214
|
String epi2String = rs.getString(RedListUtil.EPI2);
|
|
215
|
String epi3String = rs.getString(RedListUtil.EPI3);
|
|
216
|
String gueltString = rs.getString(RedListUtil.GUELT);
|
|
217
|
String trivialString = rs.getString(RedListUtil.TRIVIAL);
|
|
218
|
String authorBasiString = rs.getString(RedListUtil.AUTOR_BASI);
|
|
219
|
String hybString = rs.getString(RedListUtil.HYB);
|
|
220
|
String florString = rs.getString(RedListUtil.FLOR);
|
|
221
|
String atlasIdxString = rs.getString(RedListUtil.ATLAS_IDX);
|
|
222
|
String kartString = rs.getString(RedListUtil.KART);
|
|
223
|
String rl2015String = rs.getString(RedListUtil.RL2015);
|
|
224
|
String ehrdString = rs.getString(RedListUtil.EHRD);
|
|
225
|
String wisskString = rs.getString(RedListUtil.WISSK);
|
|
226
|
|
|
227
|
TaxonBase<?> taxonBase = null;
|
|
228
|
if(authorBasiString.trim().contains(RedListUtil.AUCT)){
|
|
229
|
taxonBase = Taxon.NewInstance(name, null);
|
|
230
|
taxonBase.setAppendedPhrase(RedListUtil.AUCT);
|
|
231
|
}
|
|
232
|
else if(gueltString.equals(RedListUtil.GUELT_ACCEPTED_TAXON)){
|
|
233
|
taxonBase = Taxon.NewInstance(name, null);
|
|
234
|
}
|
|
235
|
else if(gueltString.equals(RedListUtil.GUELT_SYNONYM) || gueltString.equals(RedListUtil.GUELT_BASIONYM)){
|
|
236
|
taxonBase = Synonym.NewInstance(name, null);
|
|
237
|
}
|
|
238
|
else{
|
|
239
|
RedListUtil.logMessage(id, "Taxon was not created!! Unknown value for "+RedListUtil.GUELT+"!", logger);
|
|
240
|
return null;
|
|
241
|
}
|
|
242
|
|
|
243
|
//common name
|
|
244
|
if(taxonBase.isInstanceOf(Taxon.class) && trivialString!=null){
|
|
245
|
Taxon taxon = HibernateProxyHelper.deproxy(taxonBase, Taxon.class);
|
|
246
|
TaxonDescription description = TaxonDescription.NewInstance(taxon);
|
|
247
|
description.addElement(CommonTaxonName.NewInstance(trivialString, Language.GERMAN()));
|
|
248
|
}
|
|
249
|
|
|
250
|
//add annotations
|
|
251
|
taxonBase.addExtension(florString, extensionTypeFlor);
|
|
252
|
taxonBase.addExtension(atlasIdxString, extensionTypeAtlasIdx);
|
|
253
|
taxonBase.addExtension(kartString, extensionTypeKart);
|
|
254
|
taxonBase.addExtension(rl2015String, extensionTypeRl2015);
|
|
255
|
taxonBase.addExtension(ehrdString, extensionTypeEhrd);
|
|
256
|
taxonBase.addExtension(wisskString, extensionTypeWissk);
|
|
257
|
|
|
258
|
//check taxon name consistency
|
|
259
|
checkTaxonConsistency(id, taxNameString, hybString, epi1String, epi2String, epi3String, taxonBase, state);
|
|
260
|
return taxonBase;
|
|
261
|
}
|
|
262
|
|
|
263
|
private void importAuthors(RedListGefaesspflanzenImportState state, ResultSet rs, INonViralName name) throws SQLException {
|
|
264
|
|
|
265
|
long id = rs.getLong(RedListUtil.NAMNR);
|
|
266
|
String nomZusatzString = rs.getString(RedListUtil.NOM_ZUSATZ);
|
|
267
|
String taxZusatzString = rs.getString(RedListUtil.TAX_ZUSATZ);
|
|
268
|
String zusatzString = rs.getString(RedListUtil.ZUSATZ);
|
|
269
|
String authorKombString = rs.getString(RedListUtil.AUTOR_KOMB);
|
|
270
|
String authorBasiString = rs.getString(RedListUtil.AUTOR_BASI);
|
|
271
|
String hybString = rs.getString(RedListUtil.HYB);
|
|
272
|
|
|
273
|
//combination author
|
|
274
|
if(authorKombString.contains(RedListUtil.EX)){
|
|
275
|
// multiple ex authors will be reduced to only the last one
|
|
276
|
// e.g. Almq. ex Sternström ex Dahlst. -> Almq. ex Dahlst.
|
|
277
|
//first author is ex combination author
|
|
278
|
String exAuthorString = RedListUtil.getExAuthorOfExAuthorshipString(authorKombString);
|
|
279
|
TeamOrPersonBase<?> exAuthor = (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, exAuthorString);
|
|
280
|
name.setExCombinationAuthorship(exAuthor);
|
|
281
|
//the last author is the combination author
|
|
282
|
String authorString = RedListUtil.getAuthorOfExAuthorshipString(authorKombString);
|
|
283
|
TeamOrPersonBase<?> combAuthor = (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, authorString);
|
|
284
|
name.setCombinationAuthorship(combAuthor);
|
|
285
|
}
|
|
286
|
else if(authorKombString.trim().contains(RedListUtil.AUCT)){
|
|
287
|
RedListUtil.logMessage(id, "AUCT information in "+RedListUtil.AUTOR_KOMB+" column", logger);
|
|
288
|
}
|
|
289
|
else if(CdmUtils.isNotBlank(authorKombString)){
|
|
290
|
TeamOrPersonBase<?> authorKomb = (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, authorKombString);
|
|
291
|
name.setCombinationAuthorship(authorKomb);
|
|
292
|
}
|
|
293
|
//basionym author
|
|
294
|
if(authorBasiString.contains(RedListUtil.EX)){
|
|
295
|
TeamOrPersonBase<?> authorExBasi= (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, RedListUtil.getExAuthorOfExAuthorshipString(authorBasiString));
|
|
296
|
if(CdmUtils.isBlank(authorKombString)){
|
|
297
|
name.setExCombinationAuthorship(authorExBasi);
|
|
298
|
}
|
|
299
|
else{
|
|
300
|
name.setExBasionymAuthorship(authorExBasi);
|
|
301
|
}
|
|
302
|
TeamOrPersonBase<?> authorBasi= (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, RedListUtil.getAuthorOfExAuthorshipString(authorBasiString));
|
|
303
|
if(CdmUtils.isBlank(authorKombString)){
|
|
304
|
name.setCombinationAuthorship(authorBasi);
|
|
305
|
}
|
|
306
|
else{
|
|
307
|
name.setBasionymAuthorship(authorBasi);
|
|
308
|
}
|
|
309
|
}
|
|
310
|
else if(CdmUtils.isNotBlank(authorBasiString)){
|
|
311
|
//this seems to be a convention in the source database: When there is only a single author then only the "AUTOR_BASI" column is used
|
|
312
|
TeamOrPersonBase<?> authorBasi= (TeamOrPersonBase<?>) state.getRelatedObject(RedListUtil.AUTHOR_NAMESPACE, authorBasiString);
|
|
313
|
if(CdmUtils.isBlank(authorKombString)){
|
|
314
|
name.setCombinationAuthorship(authorBasi);
|
|
315
|
}
|
|
316
|
else{
|
|
317
|
name.setBasionymAuthorship(authorBasi);
|
|
318
|
}
|
|
319
|
}
|
|
320
|
|
|
321
|
//check authorship consistency
|
|
322
|
String authorString = rs.getString(RedListUtil.AUTOR);
|
|
323
|
checkNameConsistency(id, nomZusatzString, taxZusatzString, zusatzString, authorString, hybString, name);
|
|
324
|
}
|
|
325
|
|
|
326
|
private INonViralName importName(RedListGefaesspflanzenImportState state, ResultSet rs, Set<ITaxonNameBase> namesToSave) throws SQLException {
|
|
327
|
|
|
328
|
long id = rs.getLong(RedListUtil.NAMNR);
|
|
329
|
String taxNameString = rs.getString(RedListUtil.TAXNAME);
|
|
330
|
String rangString = rs.getString(RedListUtil.RANG);
|
|
331
|
String ep1String = rs.getString(RedListUtil.EPI1);
|
|
332
|
String ep2String = rs.getString(RedListUtil.EPI2);
|
|
333
|
String ep3String = rs.getString(RedListUtil.EPI3);
|
|
334
|
String nomZusatzString = rs.getString(RedListUtil.NOM_ZUSATZ);
|
|
335
|
String hybString = rs.getString(RedListUtil.HYB);
|
|
336
|
String formelString = rs.getString(RedListUtil.FORMEL);
|
|
337
|
|
|
338
|
if(CdmUtils.isBlank(taxNameString) && CdmUtils.isBlank(ep1String)){
|
|
339
|
RedListUtil.logMessage(id, "No name found!", logger);
|
|
340
|
}
|
|
341
|
|
|
342
|
INonViralName name = null;
|
|
343
|
Rank rank = makeRank(id, state, rangString, CdmUtils.isNotBlank(ep3String));
|
|
344
|
//cultivar
|
|
345
|
if(rank!= null && rank.equals(Rank.CULTIVAR())){
|
|
346
|
ICultivarPlantName cultivar = TaxonNameFactory.NewCultivarInstance(rank);
|
|
347
|
cultivar.setGenusOrUninomial(ep1String);
|
|
348
|
cultivar.setSpecificEpithet(ep2String);
|
|
349
|
cultivar.setCultivarName(ep3String);
|
|
350
|
name = cultivar;
|
|
351
|
}
|
|
352
|
//botanical names
|
|
353
|
else{
|
|
354
|
name = TaxonNameFactory.NewBotanicalInstance(rank);
|
|
355
|
|
|
356
|
//ep1 should always be present
|
|
357
|
if(CdmUtils.isBlank(ep1String)){
|
|
358
|
RedListUtil.logMessage(id, RedListUtil.EPI1+" is empty!", logger);
|
|
359
|
}
|
|
360
|
name.setGenusOrUninomial(ep1String);
|
|
361
|
if(CdmUtils.isNotBlank(ep2String)){
|
|
362
|
if(rank!=null && rank.isInfraGenericButNotSpeciesGroup()){
|
|
363
|
name.setInfraGenericEpithet(ep2String);
|
|
364
|
}
|
|
365
|
else{
|
|
366
|
name.setSpecificEpithet(ep2String);
|
|
367
|
}
|
|
368
|
}
|
|
369
|
if(CdmUtils.isNotBlank(ep3String)){
|
|
370
|
name.setInfraSpecificEpithet(ep3String);
|
|
371
|
}
|
|
372
|
|
|
373
|
|
|
374
|
//nomenclatural status
|
|
375
|
if(CdmUtils.isNotBlank(nomZusatzString)){
|
|
376
|
NomenclaturalStatusType statusType = makeNomenclaturalStatus(id, state, nomZusatzString);
|
|
377
|
if(statusType!=null){
|
|
378
|
NomenclaturalStatus status = NomenclaturalStatus.NewInstance(statusType);
|
|
379
|
//special case for invalid names where the DB entry contains
|
|
380
|
//additional information in brackets e.g. "nom. inval. (sine basion.)"
|
|
381
|
if(statusType.equals(NomenclaturalStatusType.INVALID()) || statusType.equals(NomenclaturalStatusType.REJECTED()) ){
|
|
382
|
Pattern pattern = Pattern.compile("\\((.*?)\\)");
|
|
383
|
Matcher matcher = pattern.matcher(nomZusatzString);
|
|
384
|
if (matcher.find()){
|
|
385
|
status.setRuleConsidered(matcher.group(1));
|
|
386
|
}
|
|
387
|
}
|
|
388
|
name.addStatus(status);
|
|
389
|
}
|
|
390
|
}
|
|
391
|
//hybrid
|
|
392
|
if(CdmUtils.isNotBlank(hybString)){
|
|
393
|
//more than two hybrids not yet handled by name parser
|
|
394
|
//TODO: use parser when implemented to fully support hybrids
|
|
395
|
if(taxNameString.split(RedListUtil.HYB_SIGN).length>2){
|
|
396
|
name = TaxonNameFactory.NewBotanicalInstance(rank);
|
|
397
|
name.setTitleCache(taxNameString, true);
|
|
398
|
}
|
|
399
|
else if(hybString.equals(RedListUtil.HYB_X)){
|
|
400
|
name.setBinomHybrid(true);
|
|
401
|
}
|
|
402
|
else if(hybString.equals(RedListUtil.HYB_G)){
|
|
403
|
name.setMonomHybrid(true);
|
|
404
|
}
|
|
405
|
else if(hybString.equals(RedListUtil.HYB_XF) || hybString.equals(RedListUtil.HYB_XU)){
|
|
406
|
name.setHybridFormula(true);
|
|
407
|
String fullFormula = buildHybridFormula(ep1String, ep2String, ep3String, rank);
|
|
408
|
name = NonViralNameParserImpl.NewInstance().parseFullName(fullFormula, NomenclaturalCode.ICNAFP, rank);
|
|
409
|
}
|
|
410
|
else if(hybString.equals(RedListUtil.HYB_N)){
|
|
411
|
name = NonViralNameParserImpl.NewInstance().parseFullName(taxNameString, NomenclaturalCode.ICNAFP, rank);
|
|
412
|
}
|
|
413
|
else if(hybString.equals(RedListUtil.HYB_GF)){
|
|
414
|
if(ep1String.contains(RedListUtil.HYB_SIGN)){
|
|
415
|
name = NonViralNameParserImpl.NewInstance().parseFullName(ep1String, NomenclaturalCode.ICNAFP, rank);
|
|
416
|
}
|
|
417
|
else{
|
|
418
|
RedListUtil.logMessage(id, "HYB is "+hybString+" but "+RedListUtil.HYB+" does not contain "+RedListUtil.HYB_SIGN, logger);
|
|
419
|
}
|
|
420
|
}
|
|
421
|
else if(hybString.equals(RedListUtil.HYB_XS)){
|
|
422
|
//nothing to do
|
|
423
|
}
|
|
424
|
else{
|
|
425
|
logger.error("HYB value "+hybString+" not yet handled");
|
|
426
|
}
|
|
427
|
//save hybrid formula
|
|
428
|
if(CdmUtils.isNotBlank(formelString)){
|
|
429
|
Annotation annotation = Annotation.NewDefaultLanguageInstance(formelString);
|
|
430
|
annotation.setAnnotationType(AnnotationType.TECHNICAL());
|
|
431
|
name.addAnnotation(annotation);
|
|
432
|
}
|
|
433
|
}
|
|
434
|
}
|
|
435
|
|
|
436
|
//add source
|
|
437
|
ImportHelper.setOriginalSource(name, state.getTransactionalSourceReference(), id, RedListUtil.NAME_NAMESPACE);
|
|
438
|
|
|
439
|
namesToSave.add(name);
|
|
440
|
return name;
|
|
441
|
}
|
|
442
|
|
|
443
|
private String buildHybridFormula(String ep1String, String ep2String, String ep3String, Rank rank) {
|
|
444
|
String fullFormula = null;
|
|
445
|
if(ep1String.contains(RedListUtil.HYB_SIGN)){
|
|
446
|
fullFormula = ep1String;
|
|
447
|
}
|
|
448
|
else if(ep2String.contains(RedListUtil.HYB_SIGN)){
|
|
449
|
String[] split = ep2String.split(RedListUtil.HYB_SIGN);
|
|
450
|
String hybridFormula1 = ep1String+" "+split[0].trim();
|
|
451
|
String hybridFormula2 = ep1String+" "+split[1].trim();
|
|
452
|
//check if the genus is mentioned in EP2 or not
|
|
453
|
String[] secondHybrid = split[1].trim().split(" ");
|
|
454
|
//check if the genus is abbreviated like e.g. Centaurea jacea × C. decipiens
|
|
455
|
if(secondHybrid.length>1 && secondHybrid[0].matches("[A-Z]\\.")){
|
|
456
|
hybridFormula2 = ep1String+" "+split[1].trim().substring(3);
|
|
457
|
}
|
|
458
|
else if(secondHybrid.length>1 && secondHybrid[0].matches("[A-Z].*")){
|
|
459
|
hybridFormula2 = split[1].trim();
|
|
460
|
}
|
|
461
|
if(CdmUtils.isNotBlank(ep3String)){
|
|
462
|
hybridFormula1 += " "+rank.getAbbreviation()+" "+ep3String;
|
|
463
|
hybridFormula2 += " "+rank.getAbbreviation()+" "+ep3String;
|
|
464
|
}
|
|
465
|
fullFormula = hybridFormula1+" "+RedListUtil.HYB_SIGN+" "+hybridFormula2;
|
|
466
|
}
|
|
467
|
else if(ep3String.contains(RedListUtil.HYB_SIGN)){
|
|
468
|
String[] split = ep3String.split(RedListUtil.HYB_SIGN);
|
|
469
|
String hybridFormula1 = ep1String+" "+ep2String+" "+rank.getAbbreviation()+" "+split[0].trim();
|
|
470
|
String hybridFormula2 = ep1String+" "+ep2String+" "+rank.getAbbreviation()+" "+split[1].trim();
|
|
471
|
//check if the genus is mentioned in EP3 or not
|
|
472
|
String[] secondHybrid = split[1].trim().split(" ");
|
|
473
|
//check if the genus is abbreviated like e.g. Centaurea jacea jacea × C. jacea subsp. decipiens
|
|
474
|
if(secondHybrid.length>1 && secondHybrid[0].matches("[A-Z]\\.")){
|
|
475
|
hybridFormula2 = ep1String+" "+split[1].trim().substring(3);
|
|
476
|
}
|
|
477
|
else if(secondHybrid.length>1 && secondHybrid[0].matches("[A-Z].*")){
|
|
478
|
hybridFormula2 = split[1].trim();
|
|
479
|
}
|
|
480
|
fullFormula = hybridFormula1+" "+RedListUtil.HYB_SIGN+" "+hybridFormula2;
|
|
481
|
}
|
|
482
|
return fullFormula;
|
|
483
|
}
|
|
484
|
|
|
485
|
private void checkNameConsistency(long id, String nomZusatzString, String taxZusatzString,
|
|
486
|
String zusatzString, String authorString, String hybString, INonViralName name) {
|
|
487
|
String authorshipCache = name.getAuthorshipCache();
|
|
488
|
//FIXME: remove split length check when name parser can parse multiple hybrid parents
|
|
489
|
if(hybString.equals(RedListUtil.HYB_XF) && name.getTitleCache().split(RedListUtil.HYB_SIGN).length==2){
|
|
490
|
if(name.getHybridChildRelations().isEmpty()){
|
|
491
|
RedListUtil.logMessage(id, "Hybrid formula but no hybrid child relations: "+name.getTitleCache(), logger);
|
|
492
|
return;
|
|
493
|
}
|
|
494
|
return;
|
|
495
|
}
|
|
496
|
|
|
497
|
if(CdmUtils.isNotBlank(zusatzString)){
|
|
498
|
authorString = authorString.replace(", "+zusatzString, "");
|
|
499
|
}
|
|
500
|
if(CdmUtils.isNotBlank(nomZusatzString)){
|
|
501
|
authorString = authorString.replace(", "+nomZusatzString, "");
|
|
502
|
}
|
|
503
|
if(CdmUtils.isNotBlank(taxZusatzString)){
|
|
504
|
authorString = authorString.replace(", "+taxZusatzString, "");
|
|
505
|
}
|
|
506
|
if(authorString.equals(RedListUtil.AUCT)){
|
|
507
|
authorString = "";
|
|
508
|
}
|
|
509
|
if(!STRICT_TITLE_CHECK && authorString.matches(".*ex.*ex.*")){
|
|
510
|
return;
|
|
511
|
}
|
|
512
|
if(STRICT_TITLE_CHECK){
|
|
513
|
if(!authorString.equals(authorshipCache)){
|
|
514
|
RedListUtil.logMessage(id, "Authorship inconsistent! name.authorhshipCache <-> Column "+RedListUtil.AUTOR+": "+authorshipCache+" <-> "+authorString, logger);
|
|
515
|
}
|
|
516
|
}
|
|
517
|
else{
|
|
518
|
if(CdmUtils.isNotBlank(authorString) && !authorString.startsWith(authorshipCache)){
|
|
519
|
RedListUtil.logMessage(id, "Authorship inconsistent! name.authorhshipCache <-> Column "+RedListUtil.AUTOR+": "+authorshipCache+" <-> "+authorString, logger);
|
|
520
|
}
|
|
521
|
}
|
|
522
|
}
|
|
523
|
|
|
524
|
private void checkTaxonConsistency(long id, String taxNameString, String hybString, String epi1String, String epi2String, String epi3String, TaxonBase<?> taxonBase, RedListGefaesspflanzenImportState state) {
|
|
525
|
if(taxNameString.split(RedListUtil.HYB_SIGN).length>2){
|
|
526
|
RedListUtil.logInfoMessage(id, "multiple hybrid signs. No name check for "+taxNameString, logger);
|
|
527
|
return;
|
|
528
|
}
|
|
529
|
|
|
530
|
String nameCache = taxonBase.getName().getNameCache().trim();
|
|
531
|
taxNameString = taxNameString.trim();
|
|
532
|
taxNameString = taxNameString.replaceAll(" +", " ");
|
|
533
|
|
|
534
|
|
|
535
|
if((hybString.equals(RedListUtil.HYB_X) || hybString.equals(RedListUtil.HYB_N))
|
|
536
|
&& nameCache.matches(".*\\s"+RedListUtil.HYB_SIGN+"\\w.*")){
|
|
537
|
taxNameString = taxNameString.replace(" "+RedListUtil.HYB_SIGN+" ", " "+RedListUtil.HYB_SIGN);//hybrid sign has no space after it in titleCache for binomial hybrids
|
|
538
|
taxNameString = taxNameString.replace(" x ", " "+RedListUtil.HYB_SIGN);//in some cases a standard 'x' is used
|
|
539
|
}
|
|
540
|
else if(hybString.equals(RedListUtil.HYB_G)){
|
|
541
|
taxNameString = taxNameString.replace("X ", RedListUtil.HYB_SIGN);
|
|
542
|
}
|
|
543
|
else if(hybString.equals(RedListUtil.HYB_GF)){
|
|
544
|
taxNameString = taxNameString.replace(" "+RedListUtil.HYB_SIGN+" ", " "+RedListUtil.HYB_SIGN);
|
|
545
|
}
|
|
546
|
else if(hybString.equals(RedListUtil.HYB_XF)){
|
|
547
|
nameCache = taxonBase.getName().getTitleCache();
|
|
548
|
if(nameCache.contains("sec")){
|
|
549
|
nameCache = nameCache.substring(0, nameCache.indexOf("sec"));
|
|
550
|
}
|
|
551
|
if(!STRICT_TITLE_CHECK){
|
|
552
|
taxNameString = buildHybridFormula(epi1String, epi2String, epi3String, taxonBase.getName().getRank());
|
|
553
|
}
|
|
554
|
if(taxNameString.split(RedListUtil.HYB_SIGN).length==1){
|
|
555
|
taxNameString = taxNameString.replace(RedListUtil.HYB_SIGN+" ", RedListUtil.HYB_SIGN);
|
|
556
|
}
|
|
557
|
}
|
|
558
|
|
|
559
|
if(taxNameString.endsWith("- Gruppe")){
|
|
560
|
taxNameString = taxNameString.replaceAll("- Gruppe", "species group");
|
|
561
|
}
|
|
562
|
if(taxNameString.endsWith("- group")){
|
|
563
|
taxNameString = taxNameString.replaceAll("- group", "species group");
|
|
564
|
}
|
|
565
|
|
|
566
|
taxNameString = taxNameString.replace("agg.", "aggr.");
|
|
567
|
taxNameString = taxNameString.replace("[ranglos]", "[unranked]");
|
|
568
|
|
|
569
|
if(taxonBase.getName().getRank()!=null){
|
|
570
|
if(taxonBase.getName().getRank().equals(Rank.PROLES())){
|
|
571
|
taxNameString = taxNameString.replace("proles", "prol.");
|
|
572
|
}
|
|
573
|
else if(taxonBase.getName().getRank().equals(state.getRank(RedListUtil.uuidRankCollectionSpecies))){
|
|
574
|
taxNameString = taxNameString.replace("\"Sammelart\"", "\"Coll. Species\"");
|
|
575
|
}
|
|
576
|
}
|
|
577
|
if(STRICT_TITLE_CHECK){
|
|
578
|
if(!taxNameString.trim().equals(nameCache)){
|
|
579
|
RedListUtil.logMessage(id, "Taxon name inconsistent! taxon.nameCache <-> Column "+RedListUtil.TAXNAME+": "+nameCache+" <-> "+taxNameString, logger);
|
|
580
|
}
|
|
581
|
}
|
|
582
|
else{
|
|
583
|
if(!taxNameString.startsWith(nameCache)){
|
|
584
|
RedListUtil.logMessage(id, "Taxon name inconsistent! taxon.nameCache <-> Column "+RedListUtil.TAXNAME+": "+nameCache+" <-> "+taxNameString, logger);
|
|
585
|
}
|
|
586
|
}
|
|
587
|
}
|
|
588
|
|
|
589
|
private Rank makeRank(long id, RedListGefaesspflanzenImportState state, String rankStr, boolean hasSpecificEpithet) {
|
|
590
|
Rank rank = null;
|
|
591
|
try {
|
|
592
|
if(rankStr.equals("ORA")){
|
|
593
|
//special handling for ORA because of two possibilities
|
|
594
|
if(hasSpecificEpithet){
|
|
595
|
//re-load term because the representation was changed before
|
|
596
|
return (Rank) getTermService().load(Rank.uuidInfraspecificTaxon);
|
|
597
|
}
|
|
598
|
else{
|
|
599
|
return Rank.UNRANKED_INFRAGENERIC();
|
|
600
|
}
|
|
601
|
}
|
|
602
|
else if(rankStr.equals("SAM")){
|
|
603
|
return getRank(state, RedListUtil.uuidRankCollectionSpecies, "Collective Species", "Collective Species", "\"Coll. Species\"", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.SpeciesGroup);
|
|
604
|
}
|
|
605
|
else if(rankStr.equals("SPR")){
|
|
606
|
return getRank(state, RedListUtil.uuidRankSubproles, "Subproles", "Subproles", "subproles", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
607
|
}
|
|
608
|
else if(rankStr.equals("MOD")){
|
|
609
|
return getRank(state, RedListUtil.uuidRankModification, "Modification", "Modification", "modificatio", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
610
|
}
|
|
611
|
else if(rankStr.equals("LUS")){
|
|
612
|
return getRank(state, RedListUtil.uuidRankLusus, "Lusus", "Lusus", "lusus", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
613
|
}
|
|
614
|
else if(rankStr.equals("SPI")){
|
|
615
|
return getRank(state, RedListUtil.uuidRankSubspeciesPrincipes, "Subspecies principes", "Subspecies principes", "subsp. princ.", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
616
|
}
|
|
617
|
else if(rankStr.equals("KMB")){
|
|
618
|
return getRank(state, RedListUtil.uuidRankCombination, "Combination", "Combination", "", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
619
|
}
|
|
620
|
else if(rankStr.equals("'FO")){
|
|
621
|
return getRank(state, RedListUtil.uuidRankForme, "Forme'", "Forme'", "", (OrderedTermVocabulary<Rank>) Rank.GENUS().getVocabulary(), null, RankClass.Infraspecific);
|
|
622
|
}
|
|
623
|
else{
|
|
624
|
rank = state.getTransformer().getRankByKey(rankStr);
|
|
625
|
}
|
|
626
|
} catch (UndefinedTransformerMethodException e) {
|
|
627
|
e.printStackTrace();
|
|
628
|
}
|
|
629
|
if(rank==null){
|
|
630
|
RedListUtil.logMessage(id, rankStr+" could not be associated to a known rank.", logger);
|
|
631
|
}
|
|
632
|
return rank;
|
|
633
|
}
|
|
634
|
|
|
635
|
private NomenclaturalStatusType makeNomenclaturalStatus(long id, RedListGefaesspflanzenImportState state, String nomZusatzString) {
|
|
636
|
NomenclaturalStatusType status = null;
|
|
637
|
try {
|
|
638
|
status = state.getTransformer().getNomenclaturalStatusByKey(nomZusatzString);
|
|
639
|
} catch (UndefinedTransformerMethodException e) {
|
|
640
|
e.printStackTrace();
|
|
641
|
}
|
|
642
|
if(status==null){
|
|
643
|
RedListUtil.logMessage(id, nomZusatzString+" could not be associated to a known nomenclatural status.", logger);
|
|
644
|
}
|
|
645
|
return status;
|
|
646
|
}
|
|
647
|
|
|
648
|
|
|
649
|
|
|
650
|
@Override
|
|
651
|
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs,
|
|
652
|
RedListGefaesspflanzenImportState state) {
|
|
653
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
|
|
654
|
Map<String, AgentBase<?>> authorMap = new HashMap<String, AgentBase<?>>();
|
|
655
|
|
|
656
|
try {
|
|
657
|
while (rs.next()){
|
|
658
|
String authorKombString = rs.getString(RedListUtil.AUTOR_KOMB);
|
|
659
|
|
|
660
|
if(authorKombString.contains(RedListUtil.EX)){
|
|
661
|
String[] kombSplit = authorKombString.split(RedListUtil.EX);
|
|
662
|
for (int i = 0; i < kombSplit.length; i++) {
|
|
663
|
if(!authorMap.containsKey(kombSplit[i])){
|
|
664
|
authorMap.put(kombSplit[i], getAgentService().load(state.getAuthorMap().get(kombSplit[i])));
|
|
665
|
}
|
|
666
|
}
|
|
667
|
}
|
|
668
|
else if(CdmUtils.isNotBlank(authorKombString) && !authorMap.containsKey(authorKombString)){
|
|
669
|
authorMap.put(authorKombString, getAgentService().load(state.getAuthorMap().get(authorKombString)));
|
|
670
|
}
|
|
671
|
|
|
672
|
String authorBasiString = rs.getString(RedListUtil.AUTOR_BASI);
|
|
673
|
//basionym author
|
|
674
|
if(authorBasiString.contains(RedListUtil.EX)){
|
|
675
|
String[] basiSplit = authorBasiString.split(RedListUtil.EX);
|
|
676
|
for (int i = 0; i < basiSplit.length; i++) {
|
|
677
|
if(!authorMap.containsKey(basiSplit[i])){
|
|
678
|
authorMap.put(basiSplit[i], getAgentService().load(state.getAuthorMap().get(basiSplit[i])));
|
|
679
|
}
|
|
680
|
}
|
|
681
|
}
|
|
682
|
else if(CdmUtils.isNotBlank(authorBasiString) && !authorMap.containsKey(authorBasiString)){
|
|
683
|
authorMap.put(authorBasiString, getAgentService().load(state.getAuthorMap().get(authorBasiString)));
|
|
684
|
}
|
|
685
|
}
|
|
686
|
} catch (SQLException e) {
|
|
687
|
e.printStackTrace();
|
|
688
|
}
|
|
689
|
result.put(RedListUtil.AUTHOR_NAMESPACE, authorMap);
|
|
690
|
|
|
691
|
return result;
|
|
692
|
}
|
|
693
|
|
|
694
|
@Override
|
|
695
|
protected boolean doCheck(RedListGefaesspflanzenImportState state) {
|
|
696
|
return false;
|
|
697
|
}
|
|
698
|
|
|
699
|
@Override
|
|
700
|
protected boolean isIgnore(RedListGefaesspflanzenImportState state) {
|
|
701
|
return false;
|
|
702
|
}
|
|
703
|
|
|
704
|
}
|