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.dwca.out;
|
11
|
|
12
|
import java.io.FileNotFoundException;
|
13
|
import java.io.IOException;
|
14
|
import java.io.PrintWriter;
|
15
|
import java.io.UnsupportedEncodingException;
|
16
|
import java.util.List;
|
17
|
import java.util.Set;
|
18
|
|
19
|
import org.apache.commons.lang.StringUtils;
|
20
|
import org.apache.log4j.Logger;
|
21
|
import org.springframework.stereotype.Component;
|
22
|
import org.springframework.transaction.TransactionStatus;
|
23
|
|
24
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
25
|
import eu.etaxonomy.cdm.model.common.Annotation;
|
26
|
import eu.etaxonomy.cdm.model.common.AnnotationType;
|
27
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
28
|
import eu.etaxonomy.cdm.model.common.RelationshipTermBase;
|
29
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
30
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
31
|
import eu.etaxonomy.cdm.model.name.Rank;
|
32
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
33
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
34
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
35
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
36
|
import eu.etaxonomy.cdm.model.taxon.SynonymRelationship;
|
37
|
import eu.etaxonomy.cdm.model.taxon.SynonymRelationshipType;
|
38
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
39
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
40
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
41
|
import eu.etaxonomy.cdm.model.taxon.TaxonRelationshipType;
|
42
|
|
43
|
/**
|
44
|
* @author a.mueller
|
45
|
* @created 18.04.2011
|
46
|
*/
|
47
|
@Component
|
48
|
public class DwcaTaxExport extends DwcaExportBase {
|
49
|
private static final Logger logger = Logger.getLogger(DwcaTaxExport.class);
|
50
|
|
51
|
private static final String ROW_TYPE = "http://rs.tdwg.org/dwc/terms/Taxon";
|
52
|
private static final String fileName = "coreTax.txt";
|
53
|
|
54
|
/**
|
55
|
*
|
56
|
*/
|
57
|
public DwcaTaxExport() {
|
58
|
super();
|
59
|
this.ioName = this.getClass().getSimpleName();
|
60
|
}
|
61
|
|
62
|
/** Retrieves data from a CDM DB and serializes them CDM to XML.
|
63
|
* Starts with root taxa and traverses the classification to retrieve children taxa, synonyms and relationships.
|
64
|
* Taxa that are not part of the classification are not found.
|
65
|
*
|
66
|
* @param exImpConfig
|
67
|
* @param dbname
|
68
|
* @param filename
|
69
|
*/
|
70
|
@Override
|
71
|
protected boolean doInvoke(DwcaTaxExportState state){
|
72
|
DwcaTaxExportConfigurator config = state.getConfig();
|
73
|
TransactionStatus txStatus = startTransaction(true);
|
74
|
|
75
|
DwcaMetaDataRecord metaRecord = new DwcaMetaDataRecord(true, fileName, ROW_TYPE);
|
76
|
state.addMetaRecord(metaRecord);
|
77
|
|
78
|
PrintWriter writer = null;
|
79
|
try {
|
80
|
|
81
|
writer = createPrintWriter(fileName, state);
|
82
|
|
83
|
List<TaxonNode> allNodes = getAllNodes(null);
|
84
|
int i = 0;
|
85
|
for (TaxonNode node : allNodes){
|
86
|
i++;
|
87
|
Taxon taxon = CdmBase.deproxy(node.getTaxon(), Taxon.class);
|
88
|
DwcaTaxRecord record = new DwcaTaxRecord(metaRecord, config);
|
89
|
|
90
|
NonViralName<?> name = CdmBase.deproxy(taxon.getName(), NonViralName.class);
|
91
|
Taxon parent = node.getParent() == null ? null : node.getParent().getTaxon();
|
92
|
TaxonNameBase<?, ?> basionym = name.getBasionym();
|
93
|
Classification classification = node.getClassification();
|
94
|
if (! this.recordExists(taxon)){
|
95
|
handleTaxonBase(record, taxon, name, taxon, parent, basionym, classification, null, false, false);
|
96
|
record.write(writer);
|
97
|
this.addExistingRecord(taxon);
|
98
|
}
|
99
|
|
100
|
node.getClassification().getName();
|
101
|
//synonyms
|
102
|
handleSynonyms(taxon, writer, classification, metaRecord, config);
|
103
|
|
104
|
//misapplied names
|
105
|
handleMisapplication(taxon, writer, classification, metaRecord, config);
|
106
|
|
107
|
writer.flush();
|
108
|
|
109
|
}
|
110
|
} catch (FileNotFoundException e) {
|
111
|
e.printStackTrace();
|
112
|
} catch (UnsupportedEncodingException e) {
|
113
|
e.printStackTrace();
|
114
|
} catch (ClassCastException e) {
|
115
|
e.printStackTrace();
|
116
|
} catch (IOException e) {
|
117
|
e.printStackTrace();
|
118
|
}
|
119
|
finally{
|
120
|
closeWriter(writer, state);
|
121
|
}
|
122
|
commitTransaction(txStatus);
|
123
|
return true;
|
124
|
|
125
|
}
|
126
|
|
127
|
private void handleSynonyms(Taxon taxon, PrintWriter writer, Classification classification, DwcaMetaDataRecord metaRecord, DwcaTaxExportConfigurator config) {
|
128
|
Set<SynonymRelationship> synRels = taxon.getSynonymRelations();
|
129
|
for (SynonymRelationship synRel :synRels ){
|
130
|
DwcaTaxRecord record = new DwcaTaxRecord(metaRecord, config);
|
131
|
Synonym synonym = synRel.getSynonym();
|
132
|
SynonymRelationshipType type = synRel.getType();
|
133
|
boolean isProParte = synRel.isProParte();
|
134
|
boolean isPartial = synRel.isPartial();
|
135
|
if (type == null){ // should not happen
|
136
|
type = SynonymRelationshipType.SYNONYM_OF();
|
137
|
}
|
138
|
NonViralName<?> name = CdmBase.deproxy(synonym.getName(), NonViralName.class);
|
139
|
//????
|
140
|
Taxon parent = null;
|
141
|
TaxonNameBase<?, ?> basionym = name.getBasionym();
|
142
|
|
143
|
if (! this.recordExists(synonym)){
|
144
|
handleTaxonBase(record, synonym, name, taxon, parent, basionym, classification, type, isProParte, isPartial);
|
145
|
record.write(writer);
|
146
|
this.addExistingRecord(synonym);
|
147
|
}
|
148
|
|
149
|
}
|
150
|
|
151
|
}
|
152
|
|
153
|
|
154
|
private void handleMisapplication(Taxon taxon, PrintWriter writer, Classification classification, DwcaMetaDataRecord metaRecord, DwcaTaxExportConfigurator config) {
|
155
|
Set<Taxon> misappliedNames = taxon.getMisappliedNames();
|
156
|
for (Taxon misappliedName : misappliedNames ){
|
157
|
DwcaTaxRecord record = new DwcaTaxRecord(metaRecord, config);
|
158
|
TaxonRelationshipType relType = TaxonRelationshipType.MISAPPLIED_NAME_FOR();
|
159
|
NonViralName<?> name = CdmBase.deproxy(misappliedName.getName(), NonViralName.class);
|
160
|
//????
|
161
|
Taxon parent = null;
|
162
|
TaxonNameBase<?, ?> basionym = name.getBasionym();
|
163
|
|
164
|
if (! this.recordExists(misappliedName)){
|
165
|
handleTaxonBase(record, misappliedName, name, taxon, parent, basionym, classification, relType, false, false);
|
166
|
record.write(writer);
|
167
|
this.addExistingRecord(misappliedName);
|
168
|
}
|
169
|
}
|
170
|
}
|
171
|
|
172
|
/**
|
173
|
* @param record
|
174
|
* @param taxonBase
|
175
|
* @param name
|
176
|
* @param acceptedTaxon
|
177
|
* @param parent
|
178
|
* @param basionym
|
179
|
* @param isPartial
|
180
|
* @param isProParte
|
181
|
* @param config
|
182
|
* @param type
|
183
|
*/
|
184
|
private void handleTaxonBase(DwcaTaxRecord record, TaxonBase<?> taxonBase, NonViralName<?> name,
|
185
|
Taxon acceptedTaxon, Taxon parent, TaxonNameBase<?, ?> basionym, Classification classification,
|
186
|
RelationshipTermBase<?> relType, boolean isProParte, boolean isPartial) {
|
187
|
record.setId(taxonBase.getId());
|
188
|
record.setUuid(taxonBase.getUuid());
|
189
|
|
190
|
//maybe wrong as according to the DwC-A documentation only resolvable ids are allowed, this differs from DwC documentation
|
191
|
record.setScientificNameId(name);
|
192
|
record.setScientificName(name.getTitleCache());
|
193
|
|
194
|
record.setAcceptedNameUsageId(acceptedTaxon.getUuid());
|
195
|
record.setAcceptedNameUsage(acceptedTaxon.getName() == null? acceptedTaxon.getTitleCache() : acceptedTaxon.getName().getTitleCache());
|
196
|
|
197
|
//parentNameUsage
|
198
|
if (parent != null){
|
199
|
record.setParentNameUsageId(parent.getUuid());
|
200
|
record.setParentNameUsage(parent.getTitleCache());
|
201
|
}
|
202
|
|
203
|
//originalNameUsage
|
204
|
// ??? - is not a name usage (concept)
|
205
|
if (basionym != null){
|
206
|
record.setOriginalNameUsageId(basionym.getUuid());
|
207
|
record.setOriginalNameUsage(basionym.getTitleCache());
|
208
|
}
|
209
|
|
210
|
//nameAccordingTo
|
211
|
Reference<?> sec = taxonBase.getSec();
|
212
|
if (sec == null){
|
213
|
String message = "There is a taxon without sec " + taxonBase.getTitleCache() + "( " + taxonBase.getId() + ")";
|
214
|
logger.warn(message);
|
215
|
}else{
|
216
|
record.setNameAccordingToId(taxonBase.getSec().getUuid());
|
217
|
record.setNameAccordingTo(taxonBase.getSec().getTitleCache());
|
218
|
}
|
219
|
|
220
|
//namePublishedIn
|
221
|
// ??? is not a nameUsage (concept)
|
222
|
if (name.getNomenclaturalReference() != null){
|
223
|
record.setNamePublishedInId(name.getNomenclaturalReference().getUuid());
|
224
|
record.setNamePublishedIn(name.getNomenclaturalReference() == null ? null : name.getNomenclaturalReference().getTitleCache());
|
225
|
}
|
226
|
|
227
|
// what is the exact difference to id and acceptedNameUsageId
|
228
|
record.setTaxonConceptId(taxonBase.getUuid());
|
229
|
|
230
|
//Classification
|
231
|
//FIXME all classification and rank specific fields are meant to represent the classification
|
232
|
//currently the information is only compiled for the exact same range but it should be compiled
|
233
|
//for all ranks above the rank of this taxon
|
234
|
//TODO we do not support this yet
|
235
|
record.setHigherClassification(null);
|
236
|
//... higher ranks
|
237
|
handleUninomialOrGenus(record, name);
|
238
|
if (name.getRank() != null && name.getRank().equals(Rank.SUBGENUS())){
|
239
|
record.setSubgenus(name.getNameCache());
|
240
|
}
|
241
|
//record.setSubgenus(name.getInfraGenericEpithet());
|
242
|
|
243
|
|
244
|
record.setSpecificEpithet(name.getSpecificEpithet());
|
245
|
record.setInfraspecificEpithet(name.getInfraSpecificEpithet());
|
246
|
|
247
|
record.setTaxonRank(name.getRank());
|
248
|
if (name.getRank() != null){
|
249
|
record.setVerbatimTaxonRank(name.getRank().getAbbreviation());
|
250
|
}else{
|
251
|
String message = "No rank available for " + name.getTitleCache() + "(" + name.getId() + ")";
|
252
|
logger.warn(message);
|
253
|
}
|
254
|
|
255
|
record.setScientificNameAuthorship(name.getAuthorshipCache());
|
256
|
|
257
|
// ??? - use for TextData common names?
|
258
|
record.setVernacularName(null);
|
259
|
|
260
|
record.setNomenclaturalCode(name.getNomenclaturalCode());
|
261
|
// ??? TODO Misapplied Names, inferred synonyms
|
262
|
handleTaxonomicStatus(record, name, relType, isProParte, isPartial);
|
263
|
handleNomStatus(record, taxonBase, name);
|
264
|
|
265
|
// TODO we need to differentiate technical
|
266
|
String taxonRemarks = "";
|
267
|
for (Annotation annotation : taxonBase.getAnnotations()){
|
268
|
if (AnnotationType.EDITORIAL().equals(annotation.getAnnotationType())){
|
269
|
taxonRemarks += CdmUtils.Nz(annotation.getText());
|
270
|
}
|
271
|
}
|
272
|
for (Annotation annotation : name.getAnnotations()){
|
273
|
if (AnnotationType.EDITORIAL().equals(annotation.getAnnotationType())){
|
274
|
taxonRemarks += CdmUtils.Nz(annotation.getText());
|
275
|
}
|
276
|
}
|
277
|
if (StringUtils.isNotBlank(taxonRemarks)){
|
278
|
record.setTaxonRemarks(taxonRemarks);
|
279
|
}
|
280
|
|
281
|
// TODO which date is needed here (taxon, name, sec, ... ?)
|
282
|
record.setModified(taxonBase.getUpdated());
|
283
|
|
284
|
// ???
|
285
|
record.setLanguage(null);
|
286
|
|
287
|
record.setRights(taxonBase.getRights());
|
288
|
|
289
|
//TODO
|
290
|
record.setRightsHolder(null);
|
291
|
record.setAccessRights(null);
|
292
|
|
293
|
//TODO currently only via default value
|
294
|
record.setBibliographicCitation(null);
|
295
|
record.setInformationWithheld(null);
|
296
|
|
297
|
record.setDatasetId(classification);
|
298
|
record.setDatasetName(classification.getTitleCache());
|
299
|
|
300
|
//TODO
|
301
|
record.setSource(null);
|
302
|
|
303
|
return;
|
304
|
}
|
305
|
|
306
|
/**
|
307
|
* @param record
|
308
|
* @param name
|
309
|
* @param type
|
310
|
* @param isPartial
|
311
|
* @param isProParte
|
312
|
*/
|
313
|
private void handleTaxonomicStatus(DwcaTaxRecord record,
|
314
|
NonViralName<?> name, RelationshipTermBase<?> type, boolean isProParte, boolean isPartial) {
|
315
|
if (type == null){
|
316
|
record.setTaxonomicStatus(name.getNomenclaturalCode().acceptedTaxonStatusLabel());
|
317
|
}else{
|
318
|
String status = name.getNomenclaturalCode().synonymStatusLabel();
|
319
|
if (type.equals(SynonymRelationshipType.HETEROTYPIC_SYNONYM_OF())){
|
320
|
status = "heterotypicSynonym";
|
321
|
}else if(type.equals(SynonymRelationshipType.HOMOTYPIC_SYNONYM_OF())){
|
322
|
status = "homotypicSynonym";
|
323
|
}else if(type.equals(TaxonRelationshipType.MISAPPLIED_NAME_FOR())){
|
324
|
status = "misapplied";
|
325
|
}
|
326
|
if (isProParte){
|
327
|
status = "proParteSynonym";
|
328
|
}else if (isPartial){
|
329
|
String message = "Partial synonym is not part of the gbif toxonomic status vocabulary";
|
330
|
logger.warn(message);
|
331
|
status = "partialSynonym";
|
332
|
}
|
333
|
|
334
|
record.setTaxonomicStatus(status);
|
335
|
}
|
336
|
}
|
337
|
|
338
|
/**
|
339
|
* @param record
|
340
|
* @param name
|
341
|
*/
|
342
|
private void handleUninomialOrGenus(DwcaTaxRecord record, NonViralName<?> name) {
|
343
|
//epethita
|
344
|
String firstEpi = name.getGenusOrUninomial();
|
345
|
if (!StringUtils.isBlank(firstEpi)){
|
346
|
Rank rank = name.getRank();
|
347
|
if (rank != null){
|
348
|
if (rank.isLower(Rank.GENUS())){
|
349
|
record.setGenus(firstEpi);
|
350
|
}else if (rank.equals(Rank.GENUS())){
|
351
|
record.setGenus(firstEpi);
|
352
|
}else if (rank.equals(Rank.KINGDOM())){
|
353
|
record.setKingdom(firstEpi);
|
354
|
}else if (rank.equals(Rank.PHYLUM())){
|
355
|
record.setPhylum(firstEpi);
|
356
|
}else if (rank.equals(Rank.CLASS())){
|
357
|
record.setClazz(firstEpi);
|
358
|
}else if (rank.equals(Rank.ORDER())){
|
359
|
record.setOrder(firstEpi);
|
360
|
}else if (rank.equals(Rank.FAMILY())){
|
361
|
record.setFamily(firstEpi);
|
362
|
}else{
|
363
|
// !!!
|
364
|
String message = "Rank not covered. Set uninomial as genus instead: " + rank.getLabel();
|
365
|
logger.warn(message);
|
366
|
// record.setGenus(firstEpi);
|
367
|
}
|
368
|
|
369
|
}
|
370
|
}
|
371
|
}
|
372
|
|
373
|
|
374
|
/**
|
375
|
* @param record
|
376
|
* @param taxon
|
377
|
* @param name
|
378
|
*/
|
379
|
private void handleNomStatus(DwcaTaxRecord record, TaxonBase<?> taxon,
|
380
|
NonViralName<?> name) {
|
381
|
int nStatus = name.getStatus().size();
|
382
|
if (nStatus > 0){
|
383
|
if (name.getStatus().size()> 1){
|
384
|
String warning = "There is more than 1 nomenclatural status ( " + name.getStatus().size()+ "): " + taxon.getTitleCache();
|
385
|
logger.warn(warning);
|
386
|
}
|
387
|
NomenclaturalStatusType status = name.getStatus().iterator().next().getType();
|
388
|
record.setNomenclaturalStatus(status);
|
389
|
}else{
|
390
|
record.setNomenclaturalStatus(null);
|
391
|
}
|
392
|
}
|
393
|
|
394
|
|
395
|
@Override
|
396
|
protected boolean doCheck(DwcaTaxExportState state) {
|
397
|
boolean result = true;
|
398
|
logger.warn("No check implemented for " + this.ioName);
|
399
|
return result;
|
400
|
}
|
401
|
|
402
|
|
403
|
@Override
|
404
|
protected boolean isIgnore(DwcaTaxExportState state) {
|
405
|
return ! state.getConfig().isDoTaxa();
|
406
|
}
|
407
|
|
408
|
}
|