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.berlinModel.in;
|
11
|
|
12
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_HETEROTYPIC_SYNONYM_OF;
|
13
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_HOMOTYPIC_SYNONYM_OF;
|
14
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_INCLUDED_IN;
|
15
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_MISAPPLIED_NAME_OF;
|
16
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PARTIAL_HETEROTYPIC_SYNONYM_OF;
|
17
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PARTIAL_HOMOTYPIC_SYNONYM_OF;
|
18
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PARTIAL_SYN_OF;
|
19
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PROPARTE_HETEROTYPIC_SYNONYM_OF;
|
20
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PROPARTE_HOMOTYPIC_SYNONYM_OF;
|
21
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_PROPARTE_SYN_OF;
|
22
|
import static eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer.TAX_REL_IS_SYNONYM_OF;
|
23
|
|
24
|
import java.sql.ResultSet;
|
25
|
import java.sql.SQLException;
|
26
|
import java.util.HashMap;
|
27
|
import java.util.HashSet;
|
28
|
import java.util.Map;
|
29
|
import java.util.Set;
|
30
|
import java.util.UUID;
|
31
|
|
32
|
import org.apache.commons.lang.StringUtils;
|
33
|
import org.apache.log4j.Logger;
|
34
|
import org.springframework.stereotype.Component;
|
35
|
import org.springframework.transaction.TransactionStatus;
|
36
|
|
37
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
38
|
import eu.etaxonomy.cdm.common.ResultWrapper;
|
39
|
import eu.etaxonomy.cdm.io.berlinModel.BerlinModelTransformer;
|
40
|
import eu.etaxonomy.cdm.io.berlinModel.in.validation.BerlinModelTaxonRelationImportValidator;
|
41
|
import eu.etaxonomy.cdm.io.common.IOValidator;
|
42
|
import eu.etaxonomy.cdm.io.common.ResultSetPartitioner;
|
43
|
import eu.etaxonomy.cdm.io.common.Source;
|
44
|
import eu.etaxonomy.cdm.model.common.AnnotatableEntity;
|
45
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
46
|
import eu.etaxonomy.cdm.model.common.IdentifiableSource;
|
47
|
import eu.etaxonomy.cdm.model.name.NameRelationshipType;
|
48
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus;
|
49
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
50
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
51
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
52
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
53
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
54
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
55
|
import eu.etaxonomy.cdm.model.taxon.SynonymType;
|
56
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
57
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
58
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
59
|
import eu.etaxonomy.cdm.model.taxon.TaxonRelationship;
|
60
|
import eu.etaxonomy.cdm.model.taxon.TaxonRelationshipType;
|
61
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
62
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
63
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
64
|
|
65
|
/**
|
66
|
* @author a.mueller
|
67
|
* @since 20.03.2008
|
68
|
*/
|
69
|
@Component
|
70
|
public class BerlinModelTaxonRelationImport extends BerlinModelImportBase {
|
71
|
private static final long serialVersionUID = -7234926279240842557L;
|
72
|
|
73
|
private static final Logger logger = Logger.getLogger(BerlinModelTaxonRelationImport.class);
|
74
|
|
75
|
public static final String TREE_NAMESPACE = "PTRefFk";
|
76
|
private static final Integer AUCT_REF_ID = 5959;
|
77
|
|
78
|
private static int modCount = 30000;
|
79
|
private static final String pluralString = "taxon relations";
|
80
|
private static final String dbTableName = "RelPTaxon";
|
81
|
|
82
|
|
83
|
public BerlinModelTaxonRelationImport(){
|
84
|
super(dbTableName, pluralString);
|
85
|
}
|
86
|
|
87
|
/**
|
88
|
* Creates a classification for each PTaxon reference which belongs to a taxon that is
|
89
|
* included at least in one <i>taxonomically included</i> relationship.
|
90
|
*
|
91
|
* @param state
|
92
|
* @throws SQLException
|
93
|
*/
|
94
|
private void makeClassifications(BerlinModelImportState state) throws SQLException{
|
95
|
logger.info("start make classification ...");
|
96
|
|
97
|
Set<String> idSet = getTreeReferenceIdSet(state);
|
98
|
|
99
|
//reference map
|
100
|
String nameSpace = BerlinModelReferenceImport.REFERENCE_NAMESPACE;
|
101
|
Class<?> cdmClass = Reference.class;
|
102
|
@SuppressWarnings("unchecked")
|
103
|
Map<String, Reference> refMap = (Map<String, Reference>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
104
|
|
105
|
String classificationName = "Classification - No Name";
|
106
|
|
107
|
ResultSet rs = state.getConfig().getSource().getResultSet(getClassificationQuery(state)) ;
|
108
|
int i = 0;
|
109
|
//for each reference
|
110
|
try {
|
111
|
//TODO handle case useSingleClassification = true && sourceSecId = null, which returns no record
|
112
|
boolean isFirst = true;
|
113
|
while (rs.next()){
|
114
|
|
115
|
try {
|
116
|
if ((i++ % modCount) == 0 && i!= 1 ){ logger.info("RelPTaxa handled: " + (i-1));}
|
117
|
|
118
|
Integer ptRefFkInt = nullSafeInt(rs,"PTRefFk");
|
119
|
String ptRefFk= String.valueOf(ptRefFkInt);
|
120
|
Reference ref = refMap.get(ptRefFk);
|
121
|
|
122
|
String refCache = rs.getString("RefCache");
|
123
|
if (isNotBlank(refCache)){
|
124
|
classificationName = refCache;
|
125
|
}
|
126
|
if (ref != null && StringUtils.isNotBlank(ref.getTitleCache())){
|
127
|
classificationName = ref.getTitleCache();
|
128
|
}
|
129
|
if (isFirst && isNotBlank(state.getConfig().getClassificationName())){
|
130
|
classificationName = state.getConfig().getClassificationName();
|
131
|
}
|
132
|
Classification tree = Classification.NewInstance(classificationName);
|
133
|
tree.setReference(ref);
|
134
|
if (i == 1 && state.getConfig().getClassificationUuid() != null){
|
135
|
tree.setUuid(state.getConfig().getClassificationUuid());
|
136
|
}
|
137
|
IdentifiableSource identifiableSource = IdentifiableSource.NewDataImportInstance(ptRefFk, TREE_NAMESPACE);
|
138
|
tree.addSource(identifiableSource);
|
139
|
|
140
|
getClassificationService().save(tree);
|
141
|
state.putClassificationUuidInt(ptRefFkInt, tree);
|
142
|
isFirst = false;
|
143
|
} catch (Exception e) {
|
144
|
logger.error("Error in BerlinModleTaxonRelationImport.makeClassifications: " + e.getMessage());
|
145
|
e.printStackTrace();
|
146
|
}
|
147
|
}
|
148
|
} catch (SQLException e) {
|
149
|
logger.error("Error in BerlinModleTaxonRelationImport.makeClassifications: " + e.getMessage());
|
150
|
throw e;
|
151
|
}
|
152
|
logger.info("end make classification ...");
|
153
|
|
154
|
return;
|
155
|
}
|
156
|
|
157
|
/**
|
158
|
* @return
|
159
|
* @throws SQLException
|
160
|
*/
|
161
|
private Set<String> getTreeReferenceIdSet(BerlinModelImportState state) throws SQLException {
|
162
|
Source source = state.getConfig().getSource();
|
163
|
Set<String> result = new HashSet<String>();
|
164
|
ResultSet rs = source.getResultSet(getClassificationQuery(state)) ;
|
165
|
while (rs.next()){
|
166
|
Object id = rs.getObject("PTRefFk");
|
167
|
result.add(String.valueOf(id));
|
168
|
}
|
169
|
return result;
|
170
|
}
|
171
|
|
172
|
private String getClassificationQuery(BerlinModelImportState state) {
|
173
|
boolean includeAllClassifications = state.getConfig().isIncludeAllNonMisappliedRelatedClassifications();
|
174
|
String strQuerySelect = "SELECT PTaxon.PTRefFk, r.RefCache ";
|
175
|
String strQueryFrom = " FROM RelPTaxon " +
|
176
|
" INNER JOIN PTaxon AS PTaxon ON RelPTaxon.PTNameFk2 = PTaxon.PTNameFk AND RelPTaxon.PTRefFk2 = PTaxon.PTRefFk " +
|
177
|
" INNER JOIN Reference r ON PTaxon.PTRefFk = r.RefId ";
|
178
|
String strQueryWhere = " WHERE (RelPTaxon.RelQualifierFk = 1) ";
|
179
|
if (includeAllClassifications){
|
180
|
strQueryWhere = " WHERE (RelPTaxon.RelQualifierFk <> 3) ";
|
181
|
}else{
|
182
|
if (state.getConfig().isUseSingleClassification()){
|
183
|
if (state.getConfig().getSourceSecId()!= null){
|
184
|
strQueryWhere += " AND PTaxon.PTRefFk = " + state.getConfig().getSourceSecId() + " ";
|
185
|
}else{
|
186
|
strQueryWhere += " AND (1=0) ";
|
187
|
}
|
188
|
}
|
189
|
}
|
190
|
|
191
|
String strQueryGroupBy = " GROUP BY PTaxon.PTRefFk, r.RefCache ";
|
192
|
String strQuery = strQuerySelect + " " + strQueryFrom + " " + strQueryWhere + " " + strQueryGroupBy;
|
193
|
|
194
|
|
195
|
if (includeAllClassifications){
|
196
|
//add otherdirection
|
197
|
strQuerySelect = "SELECT PTaxon.PTRefFk, r.RefCache ";
|
198
|
strQueryFrom = " FROM RelPTaxon rel " +
|
199
|
" INNER JOIN PTaxon AS PTaxon ON rel.PTNameFk1 = PTaxon.PTNameFk AND rel.PTRefFk1 = PTaxon.PTRefFk " +
|
200
|
" INNER JOIN Reference r ON PTaxon.PTRefFk = r.RefId ";
|
201
|
strQueryWhere =" WHERE (rel.RelQualifierFk <> 3) ";
|
202
|
String strAllQuery = strQuerySelect + " " + strQueryFrom + " " + strQueryWhere + " " + strQueryGroupBy;
|
203
|
strQuery = strQuery + " UNION " + strAllQuery;
|
204
|
}
|
205
|
|
206
|
|
207
|
|
208
|
boolean includeFlatClassifications = state.getConfig().isIncludeFlatClassifications();
|
209
|
//concepts with
|
210
|
if (includeFlatClassifications){
|
211
|
String strFlatQuery =
|
212
|
" SELECT pt.PTRefFk AS secRefFk, r.RefCache AS secRef " +
|
213
|
" FROM PTaxon AS pt LEFT OUTER JOIN " +
|
214
|
" Reference r ON pt.PTRefFk = r.RefId LEFT OUTER JOIN " +
|
215
|
" RelPTaxon rel1 ON pt.PTNameFk = rel1.PTNameFk2 AND pt.PTRefFk = rel1.PTRefFk2 LEFT OUTER JOIN " +
|
216
|
" RelPTaxon AS rel2 ON pt.PTNameFk = rel2.PTNameFk1 AND pt.PTRefFk = rel2.PTRefFk1 " +
|
217
|
" WHERE (rel2.RelQualifierFk IS NULL) AND (rel1.RelQualifierFk IS NULL) " +
|
218
|
" GROUP BY pt.PTRefFk, r.RefCache "
|
219
|
;
|
220
|
|
221
|
strQuery = strQuery + " UNION " + strFlatQuery;
|
222
|
}
|
223
|
|
224
|
|
225
|
|
226
|
if (state.getConfig().getClassificationQuery() != null){
|
227
|
strQuery = state.getConfig().getClassificationQuery();
|
228
|
}
|
229
|
return strQuery;
|
230
|
}
|
231
|
|
232
|
@Override
|
233
|
protected String getRecordQuery(BerlinModelImportConfigurator config) {
|
234
|
String strQuery =
|
235
|
" SELECT RelPTaxon.*, fromTaxon.RIdentifier as taxon1Id, toTaxon.RIdentifier as taxon2Id, toTaxon.PTRefFk as treeRefFk, fromTaxon.PTRefFk as fromRefFk, q.is_concept_relation " +
|
236
|
" FROM PTaxon as fromTaxon " +
|
237
|
" INNER JOIN RelPTaxon ON fromTaxon.PTNameFk = relPTaxon.PTNameFk1 AND fromTaxon.PTRefFk = relPTaxon.PTRefFk1 " +
|
238
|
" INNER JOIN PTaxon AS toTaxon ON RelPTaxon.PTNameFk2 = ToTaxon.PTNameFk AND RelPTaxon.PTRefFk2 = ToTaxon.PTRefFk " +
|
239
|
" INNER JOIN RelPTQualifier q ON q.RelPTQualifierId = RelPTaxon.RelQualifierFk " +
|
240
|
" WHERE RelPTaxon.RelPTaxonId IN ("+ID_LIST_TOKEN+") " +
|
241
|
" ORDER BY RelPTaxon.RelPTaxonId ";
|
242
|
return strQuery;
|
243
|
}
|
244
|
|
245
|
@Override
|
246
|
public boolean doPartition(ResultSetPartitioner partitioner, BerlinModelImportState state) {
|
247
|
boolean success = true ;
|
248
|
Set<TaxonBase> taxaToSave = new HashSet<>();
|
249
|
Map<String, TaxonBase> taxonMap = partitioner.getObjectMap(BerlinModelTaxonImport.NAMESPACE);
|
250
|
Map<Integer, Classification> classificationMap = new HashMap<>();
|
251
|
Map<String, Reference> refMap = partitioner.getObjectMap(BerlinModelReferenceImport.REFERENCE_NAMESPACE);
|
252
|
|
253
|
ResultSet rs = partitioner.getResultSet();
|
254
|
|
255
|
try{
|
256
|
int i = 0;
|
257
|
//for each reference
|
258
|
while (rs.next()){
|
259
|
|
260
|
if ((i++ % modCount) == 0 && i!= 1 ){ logger.info("RelPTaxa handled: " + (i-1));}
|
261
|
|
262
|
int relPTaxonId = rs.getInt("RelPTaxonId");
|
263
|
Integer taxon1Id = nullSafeInt(rs, "taxon1Id");
|
264
|
Integer taxon2Id = nullSafeInt(rs, "taxon2Id");
|
265
|
Integer ptRefFk1 = nullSafeInt(rs, "PTRefFk1");
|
266
|
Integer ptRefFk2 = nullSafeInt(rs, "PTRefFk2");
|
267
|
|
268
|
int relQualifierFk = -1;
|
269
|
try {
|
270
|
Integer relRefFk = nullSafeInt(rs,"relRefFk");
|
271
|
int treeRefFk = rs.getInt("treeRefFk");
|
272
|
int fromRefFk = rs.getInt("fromRefFk");
|
273
|
|
274
|
relQualifierFk = rs.getInt("relQualifierFk");
|
275
|
String notes = rs.getString("notes");
|
276
|
boolean isConceptRelationship = rs.getBoolean("is_concept_relation");
|
277
|
|
278
|
TaxonBase<?> taxon1 = taxonMap.get(String.valueOf(taxon1Id));
|
279
|
TaxonBase<?> taxon2 = taxonMap.get(String.valueOf(taxon2Id));
|
280
|
|
281
|
String refFk = String.valueOf(relRefFk);
|
282
|
Reference citation = refMap.get(refFk);
|
283
|
|
284
|
String microcitation = null; //does not exist in RelPTaxon
|
285
|
|
286
|
if (taxon2 != null && taxon1 != null){
|
287
|
if (!(taxon2 instanceof Taxon)){
|
288
|
logger.error("ToTaxon (ID = " + taxon2.getId()+ ", RIdentifier = " + taxon2Id + ") can't be casted to Taxon. RelPTaxon: " + relPTaxonId );
|
289
|
success = false;
|
290
|
continue;
|
291
|
}
|
292
|
AnnotatableEntity taxonRelationship = null;
|
293
|
Taxon toTaxon = (Taxon)taxon2;
|
294
|
if (isTaxonRelationship(relQualifierFk)){
|
295
|
if (!(taxon1 instanceof Taxon)){
|
296
|
logger.error("TaxonBase (ID = " + taxon1.getId()+ ", RIdentifier = " + taxon1Id + ") for TaxonRelation ("+relPTaxonId+") can't be casted to Taxon");
|
297
|
success = false;
|
298
|
continue;
|
299
|
}
|
300
|
if(state.getConfig().isEuroMed() && CdmUtils.nullSafeEqual(relRefFk, ptRefFk2)){
|
301
|
citation = null;
|
302
|
}
|
303
|
Taxon fromTaxon = (Taxon)taxon1;
|
304
|
if (relQualifierFk == TAX_REL_IS_INCLUDED_IN){
|
305
|
taxonRelationship = makeTaxonomicallyIncluded(state, classificationMap, treeRefFk, fromTaxon, toTaxon, citation, microcitation);
|
306
|
}else if (relQualifierFk == TAX_REL_IS_MISAPPLIED_NAME_OF){
|
307
|
boolean isProParte = "p.p.".equals(notes);
|
308
|
if (isProParte){
|
309
|
notes = null;
|
310
|
}
|
311
|
boolean isDoubtful = "?".equals(notes);
|
312
|
if (isDoubtful){
|
313
|
notes = null;
|
314
|
}
|
315
|
|
316
|
if (notes!= null && notes.startsWith("{non ") && notes.endsWith("}")){
|
317
|
notes = notes.substring(1, notes.length() - 1);
|
318
|
}
|
319
|
//handle auct. author
|
320
|
if (fromTaxon.getSec() == null || fromTaxon.getSec().getTitleCache().startsWith("auct.")){
|
321
|
String existingSecTitle = fromTaxon.getSec() == null ? null : fromTaxon.getSec().getTitleCache().trim();
|
322
|
String existingAppendedPhrase = fromTaxon.getAppendedPhrase();
|
323
|
if (fromTaxon.getSec() == null && isBlank(existingAppendedPhrase)){
|
324
|
existingAppendedPhrase = "auct.";
|
325
|
}
|
326
|
fromTaxon.setSec(null);
|
327
|
if (isNotBlank(existingAppendedPhrase) && isNotBlank(notes)){
|
328
|
logger.warn("Misapplied name has >1 MA relation with a note, RelId: " + relPTaxonId);
|
329
|
}
|
330
|
|
331
|
String newAppendedPhrase = CdmUtils.concat(", ", existingSecTitle, notes);
|
332
|
fromTaxon.setAppendedPhrase(CdmUtils.concat("; ", existingAppendedPhrase, newAppendedPhrase));
|
333
|
if (isBlank(fromTaxon.getAppendedPhrase())){
|
334
|
logger.warn("Appended phrase is empty. This is probably not correct. RelID: " + relPTaxonId);
|
335
|
}else if ("auct.".equals(fromTaxon.getAppendedPhrase())){
|
336
|
fromTaxon.setAppendedPhrase(null);
|
337
|
}
|
338
|
notes = null;
|
339
|
}else if (notes != null && notes.startsWith("non ")){
|
340
|
fromTaxon.setAppendedPhrase(CdmUtils.concat(", ", fromTaxon.getAppendedPhrase(), notes));
|
341
|
notes = null;
|
342
|
}
|
343
|
|
344
|
if (isProParte){
|
345
|
taxonRelationship = toTaxon.addProParteMisappliedName(fromTaxon, citation, microcitation);
|
346
|
}else{
|
347
|
taxonRelationship = toTaxon.addMisappliedName(fromTaxon, citation, microcitation);
|
348
|
}
|
349
|
if (isDoubtful){
|
350
|
((TaxonRelationship)taxonRelationship).setDoubtful(true);
|
351
|
}
|
352
|
}else if (relQualifierFk == TAX_REL_IS_PROPARTE_SYN_OF ||
|
353
|
//TODO homo/hetero
|
354
|
relQualifierFk == TAX_REL_IS_PROPARTE_HOMOTYPIC_SYNONYM_OF ||
|
355
|
relQualifierFk == TAX_REL_IS_PROPARTE_HETEROTYPIC_SYNONYM_OF ){
|
356
|
toTaxon.addProparteSynonym(fromTaxon, citation, microcitation);
|
357
|
}else if(relQualifierFk == TAX_REL_IS_PARTIAL_SYN_OF ||
|
358
|
//TODO homo/hetero
|
359
|
relQualifierFk == TAX_REL_IS_PARTIAL_HOMOTYPIC_SYNONYM_OF ||
|
360
|
relQualifierFk == TAX_REL_IS_PARTIAL_HETEROTYPIC_SYNONYM_OF ){
|
361
|
toTaxon.addPartialSynonym(fromTaxon, citation, microcitation);
|
362
|
}else{
|
363
|
handleAllRelatedTaxa(state, fromTaxon, classificationMap, fromRefFk);
|
364
|
handleAllRelatedTaxa(state, toTaxon, classificationMap, treeRefFk);
|
365
|
logger.warn("Unhandled taxon relationship: RelId:" + relPTaxonId + "; QualifierId: " + relQualifierFk);
|
366
|
}
|
367
|
}else if (isSynonymRelationship(relQualifierFk)){
|
368
|
if (!(taxon1 instanceof Synonym)){
|
369
|
logger.warn("Validated: Taxon (ID = " + taxon1.getId()+ ", RIdentifier = " + taxon1Id + ") can't be casted to Synonym");
|
370
|
success = false;
|
371
|
continue;
|
372
|
}
|
373
|
handleAllRelatedTaxa(state, toTaxon, classificationMap, treeRefFk);
|
374
|
Synonym synonym = (Synonym)taxon1;
|
375
|
if (synonym.getAcceptedTaxon()!= null){
|
376
|
logger.warn("RelID: " + relPTaxonId + ". Synonym ("+taxon1Id +") already has an accepted taxon. Create clone.");
|
377
|
synonym = (Synonym)synonym.clone();
|
378
|
}
|
379
|
makeSynRel(state, relQualifierFk, toTaxon, synonym, citation, microcitation);
|
380
|
|
381
|
if (relQualifierFk == TAX_REL_IS_SYNONYM_OF ||
|
382
|
relQualifierFk == TAX_REL_IS_HOMOTYPIC_SYNONYM_OF ||
|
383
|
relQualifierFk == TAX_REL_IS_HETEROTYPIC_SYNONYM_OF){
|
384
|
}else{
|
385
|
success = false;
|
386
|
logger.warn("Synonym relationship type not yet implemented: " + relQualifierFk);
|
387
|
}
|
388
|
//
|
389
|
notes = handleSynonymNotes(state, toTaxon, synonym, notes, relPTaxonId);
|
390
|
}else if (isConceptRelationship){
|
391
|
ResultWrapper<Boolean> isInverse = ResultWrapper.NewInstance(false);
|
392
|
ResultWrapper<Boolean> isDoubtful = ResultWrapper.NewInstance(false);
|
393
|
try {
|
394
|
TaxonRelationshipType relType = BerlinModelTransformer.taxonRelId2TaxonRelType(relQualifierFk, isInverse, isDoubtful);
|
395
|
|
396
|
if (! (taxon1 instanceof Taxon)){
|
397
|
success = false;
|
398
|
logger.error("TaxonBase (ID = " + taxon1.getId()+ ", RIdentifier = " + taxon1Id + ") can't be casted to Taxon");
|
399
|
}else{
|
400
|
Taxon fromTaxon = (Taxon)taxon1;
|
401
|
if (isInverse.getValue() == true){
|
402
|
Taxon tmp = fromTaxon;
|
403
|
fromTaxon = toTaxon;
|
404
|
toTaxon = tmp;
|
405
|
}
|
406
|
taxonRelationship = fromTaxon.addTaxonRelation(toTaxon, relType, citation, microcitation);
|
407
|
handleAllRelatedTaxa(state, toTaxon, classificationMap, treeRefFk);
|
408
|
handleAllRelatedTaxa(state, fromTaxon, classificationMap, fromRefFk);
|
409
|
if (isDoubtful.getValue() == true){
|
410
|
((TaxonRelationship)taxonRelationship).setDoubtful(true);
|
411
|
}
|
412
|
}
|
413
|
} catch (UnknownCdmTypeException e) {
|
414
|
logger.warn("TaxonRelationShipType " + relQualifierFk + " (conceptRelationship) not yet implemented");
|
415
|
success = false;
|
416
|
}
|
417
|
}else {
|
418
|
logger.warn("TaxonRelationShipType " + relQualifierFk + " not yet implemented: RelPTaxonId = " + relPTaxonId );
|
419
|
success = false;
|
420
|
}
|
421
|
|
422
|
if (taxonRelationship != null && isNotBlank(notes)){
|
423
|
doNotes(taxonRelationship, notes);
|
424
|
}
|
425
|
if (isNotBlank(notes)){
|
426
|
logger.warn("Notes in RelPTaxon should all be handled explicitly and should not exist as notes anymore. RelID: " + relPTaxonId + ". Note: " + notes);
|
427
|
}
|
428
|
taxaToSave.add(taxon2);
|
429
|
|
430
|
//TODO
|
431
|
//etc.
|
432
|
}else{
|
433
|
if (taxon2 != null && taxon1 == null){
|
434
|
logger.warn("First taxon ("+taxon1Id+") for RelPTaxon " + relPTaxonId + " does not exist in store. RelType: " + relQualifierFk);
|
435
|
}else if (taxon2 == null && taxon1 != null){
|
436
|
logger.warn("Second taxon ("+taxon2Id +") for RelPTaxon " + relPTaxonId + " does not exist in store. RelType: " + relQualifierFk);
|
437
|
}else{
|
438
|
logger.warn("Both taxa ("+taxon1Id+","+taxon2Id +") for RelPTaxon " + relPTaxonId + " do not exist in store. RelType: " + relQualifierFk);
|
439
|
}
|
440
|
|
441
|
success = false;
|
442
|
}
|
443
|
} catch (Exception e) {
|
444
|
logger.error("Exception occurred when trying to handle taxon relationship " + relPTaxonId + " relQualifierFK " + relQualifierFk + " (" + taxon1Id + ","+ taxon2Id + "): " + e.getMessage());
|
445
|
e.printStackTrace();
|
446
|
}
|
447
|
}
|
448
|
}catch(SQLException e){
|
449
|
throw new RuntimeException(e);
|
450
|
}
|
451
|
logger.info("Taxa to save: " + taxaToSave.size());
|
452
|
partitioner.startDoSave();
|
453
|
getTaxonService().saveOrUpdate(taxaToSave);
|
454
|
classificationMap = null;
|
455
|
taxaToSave = null;
|
456
|
|
457
|
return success;
|
458
|
}
|
459
|
|
460
|
|
461
|
/**
|
462
|
* @param toTaxon
|
463
|
* @param synonym
|
464
|
* @param notes
|
465
|
* @return
|
466
|
*/
|
467
|
private String handleSynonymNotes(BerlinModelImportState state, Taxon toTaxon, Synonym synonym, String notes, int relId) {
|
468
|
if (state.getConfig().isEuroMed() && isNotBlank(notes)){
|
469
|
notes = notes.trim();
|
470
|
if (notes.startsWith("[non ") && notes.endsWith("]")){
|
471
|
notes = notes.substring(5, notes.length()-1).trim();
|
472
|
String[] splits = notes.split(", nec ");
|
473
|
for (String split : splits){
|
474
|
String nameStr = split.replace("<i>", "").replace("</i>", "");
|
475
|
NonViralNameParserImpl parser = NonViralNameParserImpl.NewInstance();
|
476
|
TaxonName name;
|
477
|
NomenclaturalStatusType status = null;
|
478
|
if (nameStr.endsWith(", nom. rej.") || nameStr.endsWith(", nom. cons.")||nameStr.endsWith(", nom. illeg.")){
|
479
|
String statusStr = nameStr.endsWith(", nom. rej.")? ", nom. rej.":
|
480
|
nameStr.endsWith(", nom. cons.")? ", nom. cons.":
|
481
|
", nom. illeg.";
|
482
|
nameStr = nameStr.replace(statusStr, "");
|
483
|
statusStr = statusStr.replace(", ", "");
|
484
|
try {
|
485
|
status = NomenclaturalStatusType.getNomenclaturalStatusTypeByAbbreviation(statusStr, null);
|
486
|
} catch (UnknownCdmTypeException e) {
|
487
|
logger.warn("NomStatusType not recognized: "+ statusStr + ", RelId: " + relId);
|
488
|
}
|
489
|
}
|
490
|
|
491
|
if (nameStr.contains(",") || nameStr.contains(" in ") ){
|
492
|
name = parser.parseReferencedName(nameStr, state.getConfig().getNomenclaturalCode(), null);
|
493
|
}else if (nameStr.matches(".*\\s\\d{4}")){
|
494
|
String nameStr2 = nameStr.substring(0, nameStr.length() - 5).trim();
|
495
|
String yearStr = nameStr.substring(nameStr.length()-4);
|
496
|
name = (TaxonName)parser.parseFullName(nameStr2, state.getConfig().getNomenclaturalCode(), null);
|
497
|
Reference nomRef = name.getNomenclaturalReference();
|
498
|
if (nomRef == null){
|
499
|
nomRef = ReferenceFactory.newGeneric();
|
500
|
name.setNomenclaturalReference(nomRef);
|
501
|
}
|
502
|
nomRef.setDatePublished(TimePeriodParser.parseStringVerbatim(yearStr));
|
503
|
}else if (nameStr.endsWith(" 1831-1832")){
|
504
|
String nameStr2 = nameStr.substring(0, nameStr.length() - 10).trim();
|
505
|
name = (TaxonName)parser.parseFullName(nameStr2, state.getConfig().getNomenclaturalCode(), null);
|
506
|
Reference nomRef = name.getNomenclaturalReference();
|
507
|
if (nomRef == null){
|
508
|
nomRef = ReferenceFactory.newGeneric();
|
509
|
name.setNomenclaturalReference(nomRef);
|
510
|
}
|
511
|
nomRef.setDatePublished(TimePeriodParser.parseStringVerbatim("1831-1832"));
|
512
|
}else{
|
513
|
name = parser.parseReferencedName(nameStr, state.getConfig().getNomenclaturalCode(), null);
|
514
|
}
|
515
|
if (name.isProtectedTitleCache() || name.isProtectedNameCache()
|
516
|
|| name.getNomenclaturalReference() != null && (name.getNomenclaturalReference().isProtectedAbbrevTitleCache()|| name.getNomenclaturalReference().isProtectedTitleCache() )){
|
517
|
logger.warn("Blocking name for synonym relation could not be parsed: " + nameStr + ", RelId: "+ relId);
|
518
|
}
|
519
|
if (status != null){
|
520
|
name.addStatus(NomenclaturalStatus.NewInstance(status));
|
521
|
}
|
522
|
synonym.getName().addRelationshipFromName(name, NameRelationshipType.BLOCKING_NAME_FOR(), null, null, null);
|
523
|
|
524
|
getNameService().saveOrUpdate(name);
|
525
|
}
|
526
|
return null;
|
527
|
}else{
|
528
|
return notes;
|
529
|
}
|
530
|
}else{
|
531
|
return notes;
|
532
|
}
|
533
|
}
|
534
|
|
535
|
private void handleAllRelatedTaxa(BerlinModelImportState state, Taxon taxon,
|
536
|
Map<Integer, Classification> classificationMap, Integer secRefFk) {
|
537
|
if (taxon.getTaxonNodes().size() > 0){
|
538
|
return;
|
539
|
}else{
|
540
|
Classification classification = getClassificationTree(state, classificationMap, secRefFk);
|
541
|
classification.addChildTaxon(taxon, null, null);
|
542
|
}
|
543
|
}
|
544
|
|
545
|
@Override
|
546
|
protected void doInvoke(BerlinModelImportState state){
|
547
|
try {
|
548
|
makeClassifications(state);
|
549
|
super.doInvoke(state);
|
550
|
makeFlatClassificationTaxa(state);
|
551
|
return;
|
552
|
} catch (SQLException e) {
|
553
|
throw new RuntimeException(e);
|
554
|
}
|
555
|
|
556
|
}
|
557
|
|
558
|
|
559
|
private void makeFlatClassificationTaxa(BerlinModelImportState state) {
|
560
|
//Note: this part still does not use partitions
|
561
|
logger.info("Flat classifications start");
|
562
|
TransactionStatus txStatus = startTransaction();
|
563
|
if (! state.getConfig().isIncludeFlatClassifications()){
|
564
|
return;
|
565
|
}
|
566
|
String sql = " SELECT pt.PTRefFk AS secRefFk, pt.RIdentifier " +
|
567
|
" FROM PTaxon AS pt " +
|
568
|
" LEFT OUTER JOIN RelPTaxon ON pt.PTNameFk = RelPTaxon.PTNameFk2 AND pt.PTRefFk = RelPTaxon.PTRefFk2 " +
|
569
|
" LEFT OUTER JOIN RelPTaxon AS RelPTaxon_1 ON pt.PTNameFk = RelPTaxon_1.PTNameFk1 AND pt.PTRefFk = RelPTaxon_1.PTRefFk1 " +
|
570
|
" WHERE (RelPTaxon_1.RelQualifierFk IS NULL) AND (dbo.RelPTaxon.RelQualifierFk IS NULL) " +
|
571
|
" ORDER BY pt.PTRefFk " ;
|
572
|
ResultSet rs = state.getConfig().getSource().getResultSet(sql);
|
573
|
Map<Object, Map<String, ? extends CdmBase>> maps = getRelatedObjectsForFlatPartition(rs);
|
574
|
|
575
|
Map<String, TaxonBase> taxonMap = (Map<String, TaxonBase>) maps.get(BerlinModelTaxonImport.NAMESPACE);
|
576
|
Map<Integer, Classification> classificationMap = new HashMap<>();
|
577
|
|
578
|
rs = state.getConfig().getSource().getResultSet(sql);
|
579
|
try {
|
580
|
while (rs.next()){
|
581
|
Integer treeRefFk = rs.getInt("secRefFk");
|
582
|
String taxonId = rs.getString("RIdentifier");
|
583
|
Classification classification = getClassificationTree(state, classificationMap, treeRefFk);
|
584
|
TaxonBase<?> taxon = taxonMap.get(taxonId);
|
585
|
if (taxon == null){
|
586
|
String message = "TaxonBase for taxon id (%s) not found in taxonMap";
|
587
|
logger.warn(String.format(message, taxonId, taxonId));
|
588
|
}else if (taxon.isInstanceOf(Taxon.class)){
|
589
|
classification.addChildTaxon(CdmBase.deproxy(taxon, Taxon.class), null, null);
|
590
|
}else{
|
591
|
String message = "TaxonBase for taxon is not of class Taxon but %s (RIdentifier %s)";
|
592
|
logger.warn(String.format(message, taxon.getClass(), taxonId));
|
593
|
}
|
594
|
}
|
595
|
} catch (SQLException e) {
|
596
|
e.printStackTrace();
|
597
|
}
|
598
|
commitTransaction(txStatus);
|
599
|
logger.info("Flat classifications end");
|
600
|
|
601
|
}
|
602
|
|
603
|
@Override
|
604
|
protected String getIdQuery(BerlinModelImportState state) {
|
605
|
if (state.getConfig().getRelTaxaIdQuery() != null){
|
606
|
return state.getConfig().getRelTaxaIdQuery();
|
607
|
}else{
|
608
|
return super.getIdQuery(state);
|
609
|
}
|
610
|
}
|
611
|
|
612
|
@Override
|
613
|
public Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForPartition(ResultSet rs, BerlinModelImportState state) {
|
614
|
String nameSpace;
|
615
|
Class<?> cdmClass;
|
616
|
Set<String> idSet;
|
617
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
|
618
|
|
619
|
try{
|
620
|
Set<String> taxonIdSet = new HashSet<>();
|
621
|
Set<String> referenceIdSet = new HashSet<>();
|
622
|
// Set<String> classificationIdSet = new HashSet<>();
|
623
|
while (rs.next()){
|
624
|
handleForeignKey(rs, taxonIdSet, "taxon1Id");
|
625
|
handleForeignKey(rs, taxonIdSet, "taxon2Id");
|
626
|
// handleForeignKey(rs, classificationIdSet, "treeRefFk");
|
627
|
handleForeignKey(rs, referenceIdSet, "RelRefFk");
|
628
|
}
|
629
|
|
630
|
//taxon map
|
631
|
nameSpace = BerlinModelTaxonImport.NAMESPACE;
|
632
|
cdmClass = TaxonBase.class;
|
633
|
idSet = taxonIdSet;
|
634
|
Map<String, TaxonBase> taxonMap = (Map<String, TaxonBase>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
635
|
result.put(nameSpace, taxonMap);
|
636
|
|
637
|
// //tree map
|
638
|
// nameSpace = "Classification";
|
639
|
// cdmClass = Classification.class;
|
640
|
// idSet = classificationIdSet;
|
641
|
// Map<String, Classification> treeMap = (Map<String, Classification>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
642
|
// result.put(cdmClass, treeMap);
|
643
|
// Set<UUID> treeUuidSet = state
|
644
|
// getClassificationService().find(uuidSet);
|
645
|
//
|
646
|
//reference map
|
647
|
nameSpace = BerlinModelReferenceImport.REFERENCE_NAMESPACE;
|
648
|
cdmClass = Reference.class;
|
649
|
idSet = referenceIdSet;
|
650
|
Map<String, Reference> referenceMap = (Map<String, Reference>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
651
|
result.put(nameSpace, referenceMap);
|
652
|
|
653
|
} catch (SQLException e) {
|
654
|
throw new RuntimeException(e);
|
655
|
}
|
656
|
return result;
|
657
|
}
|
658
|
|
659
|
|
660
|
private Map<Object, Map<String, ? extends CdmBase>> getRelatedObjectsForFlatPartition( ResultSet rs) {
|
661
|
String nameSpace;
|
662
|
Class cdmClass;
|
663
|
Set<String> idSet;
|
664
|
Map<Object, Map<String, ? extends CdmBase>> result = new HashMap<>();
|
665
|
|
666
|
try{
|
667
|
Set<String> taxonIdSet = new HashSet<>();
|
668
|
Set<String> referenceIdSet = new HashSet<>();
|
669
|
// Set<String> classificationIdSet = new HashSet<>();
|
670
|
while (rs.next()){
|
671
|
handleForeignKey(rs, taxonIdSet, "RIdentifier");
|
672
|
// handleForeignKey(rs, classificationIdSet, "treeRefFk");
|
673
|
}
|
674
|
|
675
|
//taxon map
|
676
|
nameSpace = BerlinModelTaxonImport.NAMESPACE;
|
677
|
cdmClass = TaxonBase.class;
|
678
|
idSet = taxonIdSet;
|
679
|
Map<String, TaxonBase> taxonMap = (Map<String, TaxonBase>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
680
|
result.put(nameSpace, taxonMap);
|
681
|
|
682
|
// //tree map
|
683
|
// nameSpace = "Classification";
|
684
|
// cdmClass = Classification.class;
|
685
|
// idSet = classificationIdSet;
|
686
|
// Map<String, Classification> treeMap = (Map<String, Classification>)getCommonService().getSourcedObjectsByIdInSource(cdmClass, idSet, nameSpace);
|
687
|
// result.put(cdmClass, treeMap);
|
688
|
// Set<UUID> treeUuidSet = state
|
689
|
// getClassificationService().find(uuidSet);
|
690
|
//
|
691
|
|
692
|
} catch (SQLException e) {
|
693
|
throw new RuntimeException(e);
|
694
|
}
|
695
|
return result;
|
696
|
}
|
697
|
|
698
|
|
699
|
private void makeSynRel (BerlinModelImportState state, int relQualifierFk, Taxon toTaxon, Synonym synonym, Reference citation, String microcitation){
|
700
|
if (state.getConfig().isWarnForDifferingSynonymReference() && citation != null && !citation.equals(synonym.getSec())){
|
701
|
logger.warn("A synonym relationship citation is given and differs from synonym secundum. This can not be handled in CDM");
|
702
|
}
|
703
|
if (isNotBlank(microcitation) && !microcitation.equals(synonym.getSecMicroReference())){
|
704
|
logger.warn("A synonym relationship microcitation is given and differs from synonym secundum micro reference. This can not be handled in CDM");
|
705
|
}
|
706
|
if (relQualifierFk == TAX_REL_IS_HOMOTYPIC_SYNONYM_OF ||
|
707
|
relQualifierFk == TAX_REL_IS_PROPARTE_HOMOTYPIC_SYNONYM_OF ||
|
708
|
relQualifierFk == TAX_REL_IS_PARTIAL_HOMOTYPIC_SYNONYM_OF){
|
709
|
toTaxon.addHomotypicSynonym(synonym);
|
710
|
}else if (relQualifierFk == TAX_REL_IS_HETEROTYPIC_SYNONYM_OF ||
|
711
|
relQualifierFk == TAX_REL_IS_PROPARTE_HETEROTYPIC_SYNONYM_OF ||
|
712
|
relQualifierFk == TAX_REL_IS_PARTIAL_HETEROTYPIC_SYNONYM_OF){
|
713
|
toTaxon.addSynonym(synonym, SynonymType.HETEROTYPIC_SYNONYM_OF());
|
714
|
}else if (relQualifierFk == TAX_REL_IS_SYNONYM_OF ||
|
715
|
relQualifierFk == TAX_REL_IS_PROPARTE_SYN_OF ||
|
716
|
relQualifierFk == TAX_REL_IS_PARTIAL_SYN_OF){
|
717
|
toTaxon.addSynonym(synonym, SynonymType.SYNONYM_OF());
|
718
|
}else{
|
719
|
logger.warn("SynonymyRelationShipType could not be defined for relQualifierFk " + relQualifierFk + ". 'Unknown'-Type taken instead.");
|
720
|
toTaxon.addSynonym(synonym, SynonymType.SYNONYM_OF());
|
721
|
}
|
722
|
return;
|
723
|
|
724
|
}
|
725
|
|
726
|
private boolean isSynonymRelationship(int relQualifierFk){
|
727
|
if (relQualifierFk == TAX_REL_IS_SYNONYM_OF ||
|
728
|
relQualifierFk == TAX_REL_IS_HOMOTYPIC_SYNONYM_OF ||
|
729
|
relQualifierFk == TAX_REL_IS_HETEROTYPIC_SYNONYM_OF ||
|
730
|
relQualifierFk == TAX_REL_IS_PROPARTE_SYN_OF ||
|
731
|
relQualifierFk == TAX_REL_IS_PARTIAL_SYN_OF ||
|
732
|
relQualifierFk == TAX_REL_IS_PROPARTE_HOMOTYPIC_SYNONYM_OF ||
|
733
|
relQualifierFk == TAX_REL_IS_PROPARTE_HETEROTYPIC_SYNONYM_OF ||
|
734
|
relQualifierFk == TAX_REL_IS_PARTIAL_HOMOTYPIC_SYNONYM_OF ||
|
735
|
relQualifierFk == TAX_REL_IS_PARTIAL_HETEROTYPIC_SYNONYM_OF
|
736
|
){
|
737
|
return true;
|
738
|
}else{
|
739
|
return false;
|
740
|
}
|
741
|
}
|
742
|
|
743
|
private boolean isTaxonRelationship(int relQualifierFk){
|
744
|
if (relQualifierFk == TAX_REL_IS_INCLUDED_IN ||
|
745
|
relQualifierFk == TAX_REL_IS_MISAPPLIED_NAME_OF){
|
746
|
return true;
|
747
|
}else if (relQualifierFk == TAX_REL_IS_PROPARTE_SYN_OF ||
|
748
|
relQualifierFk == TAX_REL_IS_PARTIAL_SYN_OF ||
|
749
|
relQualifierFk == TAX_REL_IS_PROPARTE_HOMOTYPIC_SYNONYM_OF ||
|
750
|
relQualifierFk == TAX_REL_IS_PROPARTE_HETEROTYPIC_SYNONYM_OF ||
|
751
|
relQualifierFk == TAX_REL_IS_PARTIAL_HOMOTYPIC_SYNONYM_OF ||
|
752
|
relQualifierFk == TAX_REL_IS_PARTIAL_HETEROTYPIC_SYNONYM_OF
|
753
|
){
|
754
|
return true;
|
755
|
|
756
|
}else{
|
757
|
return false;
|
758
|
}
|
759
|
}
|
760
|
|
761
|
private TaxonNode makeTaxonomicallyIncluded(BerlinModelImportState state, Map<Integer, Classification> classificationMap,
|
762
|
int treeRefFk, Taxon child, Taxon parent, Reference citation, String microCitation){
|
763
|
Classification tree = getClassificationTree(state, classificationMap, treeRefFk);
|
764
|
return tree.addParentChild(parent, child, citation, microCitation);
|
765
|
}
|
766
|
|
767
|
private Classification getClassificationTree(BerlinModelImportState state, Map<Integer, Classification> classificationMap, int treeRefFk) {
|
768
|
if (state.getConfig().isUseSingleClassification()){
|
769
|
if (state.getConfig().getSourceSecId() != null){
|
770
|
treeRefFk = (Integer)state.getConfig().getSourceSecId();
|
771
|
}else{
|
772
|
treeRefFk = 1;
|
773
|
}
|
774
|
|
775
|
}
|
776
|
Classification tree = classificationMap.get(treeRefFk);
|
777
|
if (tree == null){
|
778
|
UUID treeUuid = state.getTreeUuidByIntTreeKey(treeRefFk);
|
779
|
if (treeUuid == null){
|
780
|
throw new IllegalStateException("treeUUID does not exist in state for " + treeRefFk );
|
781
|
}
|
782
|
tree = getClassificationService().find(treeUuid);
|
783
|
classificationMap.put(treeRefFk, tree);
|
784
|
}
|
785
|
if (tree == null){
|
786
|
throw new IllegalStateException("Tree for ToTaxon reference " + treeRefFk + " does not exist.");
|
787
|
}
|
788
|
return tree;
|
789
|
}
|
790
|
|
791
|
@Override
|
792
|
protected boolean doCheck(BerlinModelImportState state){
|
793
|
IOValidator<BerlinModelImportState> validator = new BerlinModelTaxonRelationImportValidator();
|
794
|
return validator.validate(state);
|
795
|
}
|
796
|
|
797
|
@Override
|
798
|
protected boolean isIgnore(BerlinModelImportState state){
|
799
|
return ! state.getConfig().isDoRelTaxa();
|
800
|
}
|
801
|
|
802
|
|
803
|
}
|