1
|
/**
|
2
|
* Copyright (C) 2017 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.tropicos.in;
|
10
|
|
11
|
import java.net.URI;
|
12
|
import java.net.URISyntaxException;
|
13
|
import java.util.Map;
|
14
|
import java.util.Set;
|
15
|
import java.util.UUID;
|
16
|
|
17
|
import org.springframework.stereotype.Component;
|
18
|
|
19
|
import eu.etaxonomy.cdm.api.service.dto.IdentifiedEntityDTO;
|
20
|
import eu.etaxonomy.cdm.api.service.pager.Pager;
|
21
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
22
|
import eu.etaxonomy.cdm.io.csv.in.CsvImportBase;
|
23
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
24
|
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
|
25
|
import eu.etaxonomy.cdm.model.common.Language;
|
26
|
import eu.etaxonomy.cdm.model.common.VerbatimTimePeriod;
|
27
|
import eu.etaxonomy.cdm.model.description.Feature;
|
28
|
import eu.etaxonomy.cdm.model.description.TaxonNameDescription;
|
29
|
import eu.etaxonomy.cdm.model.description.TextData;
|
30
|
import eu.etaxonomy.cdm.model.media.Media;
|
31
|
import eu.etaxonomy.cdm.model.name.INonViralName;
|
32
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
33
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
34
|
import eu.etaxonomy.cdm.model.reference.OriginalSourceType;
|
35
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
36
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
37
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
38
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
39
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
40
|
import eu.etaxonomy.cdm.model.term.DefinedTerm;
|
41
|
import eu.etaxonomy.cdm.persistence.query.MatchMode;
|
42
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
43
|
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl;
|
44
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
45
|
|
46
|
/**
|
47
|
* @author a.mueller
|
48
|
* @since 15.11.2017
|
49
|
*/
|
50
|
@Component
|
51
|
public class TropicosNameImport<STATE extends TropicosNameImportState>
|
52
|
extends CsvImportBase<TropicosNameImportConfigurator, STATE, TaxonName>{
|
53
|
|
54
|
private static final long serialVersionUID = -4111479364751713088L;
|
55
|
|
56
|
private static final String INPUT_FULLNAME_WITH_AUTHORS = "FullnameWithAuthors";
|
57
|
private static final String INPUT_FULL_NAME_NO_AUTHORS = "FullnameNoAuthors";
|
58
|
private static final String INPUT_SOURCE_ID = "SourceID";
|
59
|
|
60
|
private static final String OUTPUT_NAME_ID = "OutputNameID";
|
61
|
private static final String OUTPUT_HOW_MATCHED = "OutputHowMatched";
|
62
|
private static final String OUTPUT_FULL_NAME_WITH_AUTHORS = "OutputFullNameWithAuthors";
|
63
|
private static final String OUTPUT_ABBREV_TITLE = "OutputAbbreviatedTitle";
|
64
|
private static final String OUTPUT_COLLATION = "OutputCollation";
|
65
|
private static final String OUTPUT_VOLUME = "OutputVolume";
|
66
|
private static final String OUTPUT_ISSUE = "OutputIssue";
|
67
|
private static final String OUTPUT_PAGE = "OutputPage";
|
68
|
private static final String OUTPUT_TITLE_PAGE_YEAR = "OutputTitlePageYear";
|
69
|
private static final String OUTPUT_YEAR_PUBLISHED = "OutputYearPublished";
|
70
|
private static final String OUTPUT_NOM_STATUS = "OutputNomenclatureStatus";
|
71
|
private static final String OUTPUT_BHL_LINK = "OutputBHLLink";
|
72
|
private static final String OUTPUT_BATCH_ID = "OutputBatchID";
|
73
|
private static final String NOM_PUB_TYPE = "NomPubType";
|
74
|
private static final String IPNI_ID = "IPNI-ID";
|
75
|
|
76
|
|
77
|
/**
|
78
|
* {@inheritDoc}
|
79
|
*/
|
80
|
@Override
|
81
|
protected void handleSingleLine(STATE state) {
|
82
|
TaxonName name = makeName(state);
|
83
|
if (name == null){
|
84
|
return;
|
85
|
}
|
86
|
if (checkAndAddIdentifier(state, name, INPUT_SOURCE_ID,
|
87
|
state.getConfig().isAllowWfoDuplicates(), DefinedTerm.IDENTIFIER_NAME_WFO())){
|
88
|
return;
|
89
|
}
|
90
|
if (checkAndAddIdentifier(state, name, IPNI_ID,
|
91
|
state.getConfig().isAllowIpniDuplicates(), DefinedTerm.IDENTIFIER_NAME_IPNI())){
|
92
|
return;
|
93
|
}
|
94
|
if (checkAndAddIdentifier(state, name, OUTPUT_NAME_ID,
|
95
|
state.getConfig().isAllowTropicosDuplicates(), DefinedTerm.IDENTIFIER_NAME_TROPICOS())){
|
96
|
return;
|
97
|
}
|
98
|
|
99
|
Map<String, String> record = state.getCurrentRecord();
|
100
|
|
101
|
makeReference(state, name);
|
102
|
makeProtologue(state, name);
|
103
|
makeNomStatus(state, name);
|
104
|
|
105
|
if (record.get(OUTPUT_HOW_MATCHED) != null){
|
106
|
//ignore
|
107
|
}
|
108
|
if (record.get(OUTPUT_ISSUE) != null){
|
109
|
//ignore
|
110
|
}
|
111
|
if (record.get(OUTPUT_BATCH_ID) != null){
|
112
|
//ignore
|
113
|
}
|
114
|
if (record.get(OUTPUT_COLLATION) != null){
|
115
|
//ignore
|
116
|
}
|
117
|
state.getDedupHelper().replaceAuthorNamesAndNomRef(state, name);
|
118
|
|
119
|
getNameService().saveOrUpdate(name);
|
120
|
state.getResult().addNewRecords(TaxonName.class.getSimpleName(), 1);
|
121
|
|
122
|
makeTaxon(state, name);
|
123
|
|
124
|
}
|
125
|
|
126
|
|
127
|
/**
|
128
|
* @param state
|
129
|
* @param name
|
130
|
*/
|
131
|
private void makeNomStatus(STATE state, TaxonName name) {
|
132
|
String nomStatusStr = state.getCurrentRecord().get(OUTPUT_NOM_STATUS);
|
133
|
if (nomStatusStr == null || nomStatusStr.equalsIgnoreCase("No opinion")){
|
134
|
return;
|
135
|
}else{
|
136
|
NomenclaturalStatusType status = null;
|
137
|
try {
|
138
|
status = NomenclaturalStatusType.getNomenclaturalStatusTypeByLabel(nomStatusStr);
|
139
|
} catch (UnknownCdmTypeException e) {
|
140
|
try {
|
141
|
status = NomenclaturalStatusType.getNomenclaturalStatusTypeByAbbreviation(nomStatusStr, name);
|
142
|
} catch (UnknownCdmTypeException e1) {
|
143
|
//handle later
|
144
|
}
|
145
|
}
|
146
|
if (status == null){
|
147
|
String message = "Nomenclatural status '%s' not recognized.";
|
148
|
message = String.format(message, nomStatusStr);
|
149
|
state.getResult().addWarning(message, state.getRow());
|
150
|
}else{
|
151
|
name.addStatus(status, null, null);
|
152
|
}
|
153
|
}
|
154
|
}
|
155
|
|
156
|
/**
|
157
|
* @param state
|
158
|
* @param name
|
159
|
*/
|
160
|
private void makeProtologue(STATE state, TaxonName name) {
|
161
|
String bhlLink = state.getCurrentRecord().get(OUTPUT_BHL_LINK);
|
162
|
if (bhlLink == null){
|
163
|
return;
|
164
|
}
|
165
|
|
166
|
TextData textData = TextData.NewInstance(Feature.PROTOLOGUE());
|
167
|
this.getNameDescription(name, state).addElement(textData);
|
168
|
URI uri;
|
169
|
try {
|
170
|
uri = new URI(bhlLink);
|
171
|
textData.addMedia(Media.NewInstance(uri, null, null, null));
|
172
|
|
173
|
} catch (URISyntaxException e) {
|
174
|
String message = "(BHL) Link could not be recognized as valid URI. Link was not added to %s: %s";
|
175
|
message = String.format(message, name.getTitleCache(), bhlLink);
|
176
|
state.getResult().addWarning(message, state.getRow());
|
177
|
}
|
178
|
}
|
179
|
|
180
|
//TODO implementation must be improved when matching of taxon names with existing names is implemented
|
181
|
//=> the assumption that the only description is the description added by this import
|
182
|
//is wrong then, but once protologues are handled differently we don't need this anymore
|
183
|
//anyway
|
184
|
private TaxonNameDescription getNameDescription(TaxonName name, STATE state) {
|
185
|
Set<TaxonNameDescription> descriptions = name.getDescriptions();
|
186
|
if (descriptions.size()>1){
|
187
|
throw new IllegalStateException("Implementation does not yet support names with multiple descriptions");
|
188
|
}else if (descriptions.size()==1){
|
189
|
return descriptions.iterator().next();
|
190
|
}else{
|
191
|
TaxonNameDescription desc = TaxonNameDescription.NewInstance(name);
|
192
|
desc.addSource(OriginalSourceType.Import, null, "NameDescription", getTransactionalSourceReference(state), null);
|
193
|
return desc;
|
194
|
}
|
195
|
}
|
196
|
|
197
|
/**
|
198
|
* @param state
|
199
|
* @param name
|
200
|
*/
|
201
|
private void makeReference(STATE state, TaxonName name) {
|
202
|
|
203
|
|
204
|
Map<String, String> record = state.getCurrentRecord();
|
205
|
String type = record.get(NOM_PUB_TYPE);
|
206
|
String abbrevTitle = record.get(OUTPUT_ABBREV_TITLE);
|
207
|
String volume = record.get(OUTPUT_VOLUME);
|
208
|
String detail = record.get(OUTPUT_PAGE);
|
209
|
String titlePageYear = record.get(OUTPUT_TITLE_PAGE_YEAR);
|
210
|
String yearPublished = record.get(OUTPUT_YEAR_PUBLISHED);
|
211
|
if (CdmUtils.isBlank(abbrevTitle, volume, detail, titlePageYear, yearPublished)){
|
212
|
//TODO
|
213
|
// state.getResult().addInfo("No nomenclatural reference information given");
|
214
|
return;
|
215
|
}
|
216
|
|
217
|
//Create and set title + in-Reference
|
218
|
Reference reference;
|
219
|
if (type == null){
|
220
|
//TODO check against DB
|
221
|
reference = ReferenceFactory.newGeneric();
|
222
|
reference.setAbbrevTitle(abbrevTitle);
|
223
|
}else if (type.equals("A")){
|
224
|
reference = ReferenceFactory.newArticle();
|
225
|
Reference journal = ReferenceFactory.newJournal();
|
226
|
journal.setAbbrevTitle(abbrevTitle);
|
227
|
reference.setInJournal(journal);
|
228
|
}else if (type.equals("B")){
|
229
|
reference = ReferenceFactory.newBook();
|
230
|
reference.setAbbrevTitle(abbrevTitle);
|
231
|
}else{
|
232
|
String message = "Value for %s not recognized. Use generic reference instead";
|
233
|
message = String.format(message, NOM_PUB_TYPE);
|
234
|
state.getResult().addWarning(message, state.getRow());
|
235
|
reference = ReferenceFactory.newGeneric();
|
236
|
reference.setAbbrevTitle(abbrevTitle);
|
237
|
}
|
238
|
reference.setVolume(volume);
|
239
|
|
240
|
//date
|
241
|
if (titlePageYear != null){
|
242
|
if (yearPublished == null){
|
243
|
VerbatimTimePeriod tp = TimePeriodParser.parseStringVerbatim(titlePageYear);
|
244
|
reference.setDatePublished(tp);
|
245
|
}else{
|
246
|
VerbatimTimePeriod tp = TimePeriodParser.parseStringVerbatim(yearPublished);
|
247
|
reference.setDatePublished(tp);
|
248
|
}
|
249
|
}else if (yearPublished != null){
|
250
|
VerbatimTimePeriod tp = TimePeriodParser.parseStringVerbatim(yearPublished);
|
251
|
reference.setDatePublished(tp);
|
252
|
}
|
253
|
|
254
|
//add to name
|
255
|
name.setNomenclaturalReference(reference);
|
256
|
name.setNomenclaturalMicroReference(detail);
|
257
|
//author
|
258
|
if (state.getConfig().isAddAuthorsToReference()){
|
259
|
TeamOrPersonBase<?> author = name.getCombinationAuthorship();
|
260
|
if (author != null){
|
261
|
reference.setAuthorship(author);
|
262
|
}
|
263
|
}
|
264
|
|
265
|
//source
|
266
|
addSourceReference(state, reference);
|
267
|
}
|
268
|
|
269
|
/**
|
270
|
* Checks if the sourceId (WFO ID) already exists in the database.
|
271
|
* @param state
|
272
|
* @param name
|
273
|
* @param idAttr
|
274
|
* @param allowDuplicate
|
275
|
* @param identifierType
|
276
|
* @return <code>true</code> if sourceId already exists.
|
277
|
*/
|
278
|
private boolean checkAndAddIdentifier(STATE state, TaxonName name, String idAttr,
|
279
|
boolean allowDuplicate, DefinedTerm identifierType) {
|
280
|
String identifier = state.getCurrentRecord().get(idAttr);
|
281
|
if (identifier == null){
|
282
|
return false;
|
283
|
}
|
284
|
|
285
|
if (! allowDuplicate || state.getConfig().isReportDuplicateIdentifier()){
|
286
|
//TODO precompute existing per session or, at least, implement count
|
287
|
Pager<IdentifiedEntityDTO<TaxonName>> existing = getNameService().findByIdentifier(TaxonName.class, identifier, identifierType, MatchMode.EXACT, false, null, null, null);
|
288
|
if (existing.getCount() > 0){
|
289
|
//TODO make language configurable
|
290
|
Language language = Language.DEFAULT();
|
291
|
if (! allowDuplicate){
|
292
|
String message = "The name with the given identifier (%s: %s) exists already in the database. Record is not imported.";
|
293
|
message = String.format(message, identifierType.getPreferredRepresentation(Language.DEFAULT()).getText(), identifier);
|
294
|
state.getResult().addWarning(message, state.getRow());
|
295
|
return true;
|
296
|
}else{
|
297
|
String message = "The name with the given identifier (%s: %s) exists already in the database. Record is imported but maybe needs to be reviewed.";
|
298
|
message = String.format(message, identifierType.getPreferredRepresentation(language).getText(), identifier);
|
299
|
state.getResult().addWarning(message, state.getRow());
|
300
|
}
|
301
|
}
|
302
|
}
|
303
|
|
304
|
name.addIdentifier(identifier, identifierType);
|
305
|
return false;
|
306
|
}
|
307
|
|
308
|
private NonViralNameParserImpl parser = NonViralNameParserImpl.NewInstance();
|
309
|
|
310
|
/**
|
311
|
* @param state
|
312
|
* @return
|
313
|
*/
|
314
|
private TaxonName makeName(STATE state) {
|
315
|
Map<String, String> record = state.getCurrentRecord();
|
316
|
String fullNameStr = record.get(OUTPUT_FULL_NAME_WITH_AUTHORS);
|
317
|
String nameStr = record.get(INPUT_FULL_NAME_NO_AUTHORS);
|
318
|
String inputFullNameStr = record.get(INPUT_FULLNAME_WITH_AUTHORS);
|
319
|
if (inputFullNameStr != null && fullNameStr != null){
|
320
|
if (inputFullNameStr.replaceAll("\\s", "").equals(fullNameStr.replaceAll("\\s", ""))){
|
321
|
String message = "Full input (%s) and full output (%s) name are not equal. Record is, however, processed.";
|
322
|
message = String.format(message, inputFullNameStr, fullNameStr);
|
323
|
state.getResult().addWarning(message, state.getRow());
|
324
|
}
|
325
|
}else if (inputFullNameStr != null && fullNameStr == null){
|
326
|
fullNameStr = inputFullNameStr;
|
327
|
}
|
328
|
|
329
|
INonViralName name;
|
330
|
if (fullNameStr == null && nameStr == null){
|
331
|
String message = "No name given. No record will be imported.";
|
332
|
state.getResult().addWarning(message, state.getRow());
|
333
|
return null;
|
334
|
}else if (fullNameStr != null){
|
335
|
name = parser.parseFullName(fullNameStr, state.getConfig().getNomenclaturalCode(), null);
|
336
|
if (nameStr != null && !nameStr.equals(name.getNameCache())){
|
337
|
String message = "Name with authors (%s) and without authors (%s) is not consistent";
|
338
|
message = String.format(message, fullNameStr, nameStr);
|
339
|
state.getResult().addWarning(message, state.getRow());
|
340
|
}
|
341
|
}else{
|
342
|
name = parser.parseSimpleName(nameStr, state.getConfig().getNomenclaturalCode(), null);
|
343
|
}
|
344
|
|
345
|
if (name.isProtectedTitleCache() || name.isProtectedNameCache() || name.isProtectedAuthorshipCache()){
|
346
|
String message = "Name (%s) could not be fully parsed, but is processed";
|
347
|
message = String.format(message, name.getTitleCache());
|
348
|
state.getResult().addWarning(message, state.getRow());
|
349
|
}
|
350
|
addSourceReference(state, (TaxonName)name);
|
351
|
return (TaxonName)name;
|
352
|
}
|
353
|
|
354
|
/**
|
355
|
* @param state
|
356
|
* @param name
|
357
|
*/
|
358
|
private void addSourceReference(STATE state, IdentifiableEntity<?> entity) {
|
359
|
entity.addImportSource(null, null, getTransactionalSourceReference(state), "line " + state.getLine());
|
360
|
}
|
361
|
|
362
|
|
363
|
@Override
|
364
|
protected void refreshTransactionStatus(STATE state) {
|
365
|
super.refreshTransactionStatus(state);
|
366
|
}
|
367
|
|
368
|
|
369
|
/**
|
370
|
* @param state
|
371
|
* @param name
|
372
|
*/
|
373
|
private void makeTaxon(STATE state, TaxonName name) {
|
374
|
if (!state.getConfig().isCreateTaxa()){
|
375
|
return;
|
376
|
}else{
|
377
|
//or do we want to allow to define an own sec reference?
|
378
|
Reference sec = getTransactionalSourceReference(state);
|
379
|
Taxon taxon = Taxon.NewInstance(name, sec);
|
380
|
TaxonNode parentNode = getParentNode(state);
|
381
|
if (parentNode != null){
|
382
|
TaxonNode newNode = parentNode.addChildTaxon(taxon, null, null);
|
383
|
if (state.getConfig().isUnplaced()){
|
384
|
newNode.setUnplaced(true);
|
385
|
}
|
386
|
}
|
387
|
addSourceReference(state, taxon);
|
388
|
this.getTaxonService().saveOrUpdate(taxon);
|
389
|
state.getResult().addNewRecords(Taxon.class.getSimpleName(), 1);
|
390
|
|
391
|
}
|
392
|
}
|
393
|
|
394
|
|
395
|
/**
|
396
|
* Transactional save method to retrieve the parent node
|
397
|
* @param state
|
398
|
* @param sec
|
399
|
* @return
|
400
|
*/
|
401
|
protected TaxonNode getParentNode(STATE state) {
|
402
|
TaxonNode parentNode = state.getParentNode();
|
403
|
if (parentNode == null){
|
404
|
if (state.getConfig().getParentNodeUuid() != null){
|
405
|
parentNode = getTaxonNodeService().find(state.getConfig().getParentNodeUuid());
|
406
|
if (parentNode == null){
|
407
|
//node does not exist => create new classification
|
408
|
Classification classification = makeClassification(state);
|
409
|
parentNode = classification.getRootNode();
|
410
|
parentNode.setUuid(state.getConfig().getParentNodeUuid());
|
411
|
}
|
412
|
}else {
|
413
|
Classification classification = makeClassification(state);
|
414
|
state.getConfig().setParentNodeUuid(classification.getRootNode().getUuid());
|
415
|
parentNode = classification.getRootNode();
|
416
|
}
|
417
|
state.setParentNode(parentNode);
|
418
|
}
|
419
|
return parentNode;
|
420
|
}
|
421
|
|
422
|
|
423
|
/**
|
424
|
* @param state
|
425
|
* @param sec
|
426
|
* @return
|
427
|
*/
|
428
|
protected Classification makeClassification(STATE state) {
|
429
|
Reference ref = getTransactionalSourceReference(state);
|
430
|
String classificationStr = state.getConfig().getClassificationName();
|
431
|
if (isBlank(classificationStr)){
|
432
|
classificationStr = "Tropicos import " + UUID.randomUUID();
|
433
|
}
|
434
|
Classification classification = Classification.NewInstance(classificationStr, ref, Language.UNDETERMINED());
|
435
|
return classification;
|
436
|
}
|
437
|
|
438
|
|
439
|
|
440
|
}
|