1
|
/**
|
2
|
* Copyright (C) 2009 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.tcsxml.in;
|
11
|
|
12
|
import java.util.ArrayList;
|
13
|
import java.util.HashMap;
|
14
|
import java.util.HashSet;
|
15
|
import java.util.Iterator;
|
16
|
import java.util.List;
|
17
|
import java.util.Map;
|
18
|
import java.util.Set;
|
19
|
|
20
|
import org.apache.log4j.Logger;
|
21
|
import org.jdom.Attribute;
|
22
|
import org.jdom.Element;
|
23
|
import org.jdom.Namespace;
|
24
|
import org.jdom.filter.ElementFilter;
|
25
|
import org.jdom.filter.Filter;
|
26
|
import org.springframework.stereotype.Component;
|
27
|
|
28
|
import eu.etaxonomy.cdm.api.service.ITaxonService;
|
29
|
import eu.etaxonomy.cdm.common.ResultWrapper;
|
30
|
import eu.etaxonomy.cdm.common.XmlHelp;
|
31
|
import eu.etaxonomy.cdm.io.common.ICdmIO;
|
32
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
33
|
import eu.etaxonomy.cdm.io.common.MapWrapper;
|
34
|
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
|
35
|
import eu.etaxonomy.cdm.model.common.Language;
|
36
|
import eu.etaxonomy.cdm.model.description.CommonTaxonName;
|
37
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
38
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
39
|
import eu.etaxonomy.cdm.model.name.Rank;
|
40
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
41
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
42
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
43
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
44
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
45
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
46
|
|
47
|
|
48
|
/**
|
49
|
* @author a.mueller
|
50
|
*
|
51
|
*/
|
52
|
@Component
|
53
|
public class TcsXmlTaxonImport extends TcsXmlImportBase implements ICdmIO<TcsXmlImportState> {
|
54
|
private static final Logger logger = Logger.getLogger(TcsXmlTaxonImport.class);
|
55
|
|
56
|
private static int modCount = 30000;
|
57
|
|
58
|
public TcsXmlTaxonImport(){
|
59
|
super();
|
60
|
}
|
61
|
|
62
|
|
63
|
@Override
|
64
|
public boolean doCheck(TcsXmlImportState state){
|
65
|
boolean result = true;
|
66
|
logger.warn("Checking for Taxa not yet implemented");
|
67
|
//result &= checkArticlesWithoutJournal(bmiConfig);
|
68
|
//result &= checkPartOfJournal(bmiConfig);
|
69
|
|
70
|
return result;
|
71
|
}
|
72
|
|
73
|
/**
|
74
|
* Computes a list of all TaxonConcept ids (ref-attribute) that are related as synonyms
|
75
|
* @param elTaxonConceptList
|
76
|
* @param success
|
77
|
* @return
|
78
|
*/
|
79
|
private Set<String> makeSynonymIds(List<Element> elTaxonConceptList, ResultWrapper<Boolean> success){
|
80
|
//TODO use XPath
|
81
|
|
82
|
Set<String> result = new HashSet<String>();
|
83
|
|
84
|
Namespace tcsNamespace;
|
85
|
//for each taxonConcept
|
86
|
for (Element elTaxonConcept : elTaxonConceptList){
|
87
|
tcsNamespace = elTaxonConcept.getNamespace();
|
88
|
|
89
|
String childName = "TaxonRelationships";
|
90
|
boolean obligatory = false;
|
91
|
Element elTaxonRelationships = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
92
|
|
93
|
if (elTaxonRelationships == null){
|
94
|
continue;
|
95
|
}
|
96
|
String tcsElementName = "TaxonRelationship";
|
97
|
List<Element> elTaxonRelationshipList = elTaxonRelationships.getChildren(tcsElementName, tcsNamespace);
|
98
|
for (Element elTaxonRelationship : elTaxonRelationshipList){
|
99
|
|
100
|
String relationshipType = elTaxonRelationship.getAttributeValue("type");
|
101
|
if ("has synonym".equalsIgnoreCase(relationshipType)){
|
102
|
childName = "ToTaxonConcept";
|
103
|
obligatory = true;
|
104
|
Element elToTaxonConcept = XmlHelp.getSingleChildElement(success, elTaxonRelationship, childName, tcsNamespace, obligatory);
|
105
|
|
106
|
String linkType = elToTaxonConcept.getAttributeValue("linkType");
|
107
|
if (linkType == null || linkType.equals("local")){
|
108
|
String ref = elToTaxonConcept.getAttributeValue("ref");
|
109
|
result.add(ref);
|
110
|
}else{
|
111
|
logger.warn("External link types for synonym not yet implemented");
|
112
|
}
|
113
|
}
|
114
|
}
|
115
|
}
|
116
|
return result;
|
117
|
}
|
118
|
|
119
|
|
120
|
protected static final Reference unknownSec(){
|
121
|
ReferenceFactory refFactory = ReferenceFactory.newInstance();
|
122
|
Reference result = refFactory.newGeneric();
|
123
|
result.setTitleCache("UNKNOWN", true);
|
124
|
return result;
|
125
|
}
|
126
|
|
127
|
@Override
|
128
|
public void doInvoke(TcsXmlImportState state){
|
129
|
|
130
|
logger.info("start make TaxonConcepts ...");
|
131
|
MapWrapper<TaxonBase> taxonMap = (MapWrapper<TaxonBase>)state.getStore(ICdmIO.TAXON_STORE);
|
132
|
MapWrapper<TaxonNameBase<?,?>> taxonNameMap = (MapWrapper<TaxonNameBase<?,?>>)state.getStore(ICdmIO.TAXONNAME_STORE);
|
133
|
MapWrapper<Reference> referenceMap = (MapWrapper<Reference>)state.getStore(ICdmIO.REFERENCE_STORE);
|
134
|
Map<String, CommonTaxonName> commonNameMap = new HashMap<String, CommonTaxonName>();
|
135
|
|
136
|
ITaxonService taxonService = getTaxonService();
|
137
|
|
138
|
ResultWrapper<Boolean> success = ResultWrapper.NewInstance(true);
|
139
|
String childName;
|
140
|
boolean obligatory;
|
141
|
String idNamespace = "TaxonConcept";
|
142
|
|
143
|
TcsXmlImportConfigurator config = state.getConfig();
|
144
|
Element elDataSet = getDataSetElement(config);
|
145
|
Namespace tcsNamespace = config.getTcsXmlNamespace();
|
146
|
|
147
|
childName = "TaxonConcepts";
|
148
|
obligatory = false;
|
149
|
Element elTaxonConcepts = XmlHelp.getSingleChildElement(success, elDataSet, childName, tcsNamespace, obligatory);
|
150
|
|
151
|
String tcsElementName = "TaxonConcept";
|
152
|
List<Element> elTaxonConceptList = elTaxonConcepts.getChildren(tcsElementName, tcsNamespace);
|
153
|
|
154
|
Set<String> synonymIdSet = makeSynonymIds(elTaxonConceptList, success);
|
155
|
//TODO make the same for the Assertions
|
156
|
|
157
|
int i = 0;
|
158
|
|
159
|
//for each taxonConcept
|
160
|
for (Element elTaxonConcept : elTaxonConceptList){
|
161
|
if ((i++ % modCount) == 0 && i > 1){ logger.info("Taxa handled: " + (i-1));}
|
162
|
List<String> elementList = new ArrayList<String>();
|
163
|
|
164
|
//create TaxonName element
|
165
|
String strId = elTaxonConcept.getAttributeValue("id");
|
166
|
//TODO
|
167
|
String strConceptType = elTaxonConcept.getAttributeValue("type"); //original, revision, incomplete, aggregate, nominal
|
168
|
String strPrimary = elTaxonConcept.getAttributeValue("primary"); //If primary='true' the concept is the first level response to a query. If 'false' the concept may be a secondary concept linked directly or indirectly to the definition of a primary concept.
|
169
|
String strForm = elTaxonConcept.getAttributeValue("form"); //anamorph, teleomorph, hybrid
|
170
|
|
171
|
childName = "Name";
|
172
|
obligatory = true;
|
173
|
Element elName = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
174
|
if (isVernacular(success, elName)){
|
175
|
handleVernacularName(success, strId, elName, commonNameMap);
|
176
|
}else{
|
177
|
TaxonNameBase<?,?> taxonName = makeScientificName(elName, null, taxonNameMap, success);
|
178
|
elementList.add(childName.toString());
|
179
|
|
180
|
//TODO how to handle
|
181
|
childName = "Rank";
|
182
|
obligatory = false;
|
183
|
Element elRank = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
184
|
Rank rank = TcsXmlTaxonNameImport.makeRank(elRank);
|
185
|
if (rank != null){
|
186
|
logger.warn("Rank in TaxonIO not yet implemented");
|
187
|
}
|
188
|
elementList.add(childName.toString());
|
189
|
|
190
|
childName = "AccordingTo";
|
191
|
obligatory = false;
|
192
|
Element elAccordingTo = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
193
|
Reference sec = makeAccordingTo(elAccordingTo, referenceMap, success);
|
194
|
elementList.add(childName.toString());
|
195
|
// TODO may sec be null?
|
196
|
if (sec == null){
|
197
|
sec = unknownSec();
|
198
|
}
|
199
|
|
200
|
TaxonBase taxonBase;
|
201
|
if (synonymIdSet.contains(strId)){
|
202
|
taxonBase = Synonym.NewInstance(taxonName, sec);
|
203
|
}else{
|
204
|
taxonBase = Taxon.NewInstance(taxonName, sec);
|
205
|
}
|
206
|
|
207
|
childName = "TaxonRelationships";
|
208
|
obligatory = false;
|
209
|
Element elTaxonRelationships = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
210
|
makeTaxonRelationships(taxonBase, elTaxonRelationships, success);
|
211
|
elementList.add(childName.toString());
|
212
|
|
213
|
childName = "SpecimenCircumscription";
|
214
|
obligatory = false;
|
215
|
Element elSpecimenCircumscription = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
216
|
makeSpecimenCircumscription(taxonBase, elSpecimenCircumscription, success);
|
217
|
elementList.add(childName.toString());
|
218
|
|
219
|
childName = "CharacterCircumscription";
|
220
|
obligatory = false;
|
221
|
Element elCharacterCircumscription = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
222
|
makeCharacterCircumscription(taxonBase, elCharacterCircumscription, success);
|
223
|
elementList.add(childName.toString());
|
224
|
|
225
|
|
226
|
childName = "ProviderLink";
|
227
|
obligatory = false;
|
228
|
Element elProviderLink = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
229
|
makeProviderLink(taxonBase, elProviderLink, success);
|
230
|
elementList.add(childName.toString());
|
231
|
|
232
|
childName = "ProviderSpecificData";
|
233
|
obligatory = false;
|
234
|
Element elProviderSpecificData = XmlHelp.getSingleChildElement(success, elTaxonConcept, childName, tcsNamespace, obligatory);
|
235
|
makeProviderSpecificData(taxonBase, elProviderSpecificData, success);
|
236
|
elementList.add(childName.toString());
|
237
|
|
238
|
testAdditionalElements(elTaxonConcept, elementList);
|
239
|
ImportHelper.setOriginalSource(taxonBase, config.getSourceReference(), strId, idNamespace);
|
240
|
taxonMap.put(strId, taxonBase);
|
241
|
}
|
242
|
|
243
|
}
|
244
|
state.setCommonNameMap(commonNameMap);
|
245
|
|
246
|
//invokeRelations(source, cdmApp, deleteAll, taxonMap, referenceMap);
|
247
|
logger.info(i + " taxa handled. Saving ...");
|
248
|
taxonService.save(taxonMap.objects());
|
249
|
logger.info("end makeTaxa ...");
|
250
|
if (!success.getValue()){
|
251
|
state.setUnsuccessfull();
|
252
|
}
|
253
|
return;
|
254
|
}
|
255
|
|
256
|
private void handleVernacularName(ResultWrapper<Boolean> success, String taxonId, Element elName, Map<String, CommonTaxonName> commonNameMap) {
|
257
|
String name = elName.getTextNormalize();
|
258
|
//TODO ref
|
259
|
|
260
|
Language language = null;
|
261
|
String strLanguage = elName.getAttributeValue("language");
|
262
|
//TODO
|
263
|
//Language
|
264
|
if (strLanguage != null){
|
265
|
language = Language.getLanguageByLabel(strLanguage);
|
266
|
if (language == null){
|
267
|
language = Language.getLanguageByDescription(strLanguage);
|
268
|
}
|
269
|
if (language == null){
|
270
|
logger.warn("language ("+strLanguage+") not found for name " + name);
|
271
|
}
|
272
|
}
|
273
|
CommonTaxonName commonName = CommonTaxonName.NewInstance(name, language);
|
274
|
commonNameMap.put(taxonId, commonName);
|
275
|
|
276
|
//TODO check other elements
|
277
|
}
|
278
|
|
279
|
|
280
|
private boolean isVernacular(ResultWrapper<Boolean> success, Element elName) {
|
281
|
try{
|
282
|
String strScientific = elName.getAttributeValue("scientific");
|
283
|
boolean scientific = Boolean.valueOf(strScientific);
|
284
|
return ! scientific;
|
285
|
} catch (Exception e) {
|
286
|
logger.warn("Value for scientific is not boolean");
|
287
|
success.setValue(false);
|
288
|
return false;
|
289
|
}
|
290
|
}
|
291
|
|
292
|
|
293
|
|
294
|
private boolean hasIsSynonymRelation(Element taxonConcept, Namespace rdfNamespace){
|
295
|
boolean result = false;
|
296
|
if (taxonConcept == null || ! "TaxonConcept".equalsIgnoreCase(taxonConcept.getName()) ){
|
297
|
return false;
|
298
|
}
|
299
|
|
300
|
String elName = "relationshipCategory";
|
301
|
Filter filter = new ElementFilter(elName, taxonConcept.getNamespace());
|
302
|
Iterator<Element> relationshipCategories = taxonConcept.getDescendants(filter);
|
303
|
while (relationshipCategories.hasNext()){
|
304
|
Element relationshipCategory = relationshipCategories.next();
|
305
|
Attribute resource = relationshipCategory.getAttribute("resource", rdfNamespace);
|
306
|
String isSynonymFor = "http://rs.tdwg.org/ontology/voc/TaxonConcept#IsSynonymFor";
|
307
|
if (resource != null && isSynonymFor.equalsIgnoreCase(resource.getValue()) ){
|
308
|
return true;
|
309
|
}
|
310
|
}
|
311
|
return result;
|
312
|
}
|
313
|
|
314
|
|
315
|
/**
|
316
|
* @param elTaxonRelationships
|
317
|
* @param success
|
318
|
*/
|
319
|
private TaxonNameBase<?, ?> makeScientificName(Element elName, NomenclaturalCode code, MapWrapper<? extends TaxonNameBase<?,?>> objectMap, ResultWrapper<Boolean> success){
|
320
|
TaxonNameBase<?, ?> result = null;
|
321
|
if (elName != null){
|
322
|
String language = elName.getAttributeValue("language");
|
323
|
//Language
|
324
|
if (language != null){
|
325
|
logger.warn("language for name not yet implemented. Language for scientific name should always be Latin");
|
326
|
}
|
327
|
Class<? extends IdentifiableEntity> clazz = (Class<? extends IdentifiableEntity>)NonViralName.class;
|
328
|
if (code != null){
|
329
|
clazz = code.getCdmClass();
|
330
|
}
|
331
|
result = (TaxonNameBase<?,?>)makeReferenceType (elName, clazz , objectMap, success);
|
332
|
if(result == null){
|
333
|
logger.warn("Name not found");
|
334
|
success.setValue(false);
|
335
|
}
|
336
|
}else{
|
337
|
logger.warn("Name element is null");
|
338
|
}
|
339
|
return result;
|
340
|
}
|
341
|
|
342
|
|
343
|
/**
|
344
|
* @param elTaxonRelationships
|
345
|
* @param success
|
346
|
*/
|
347
|
private void makeTaxonRelationships(TaxonBase name, Element elTaxonRelationships, ResultWrapper<Boolean> success){
|
348
|
//TaxonRelationships are handled in TcsXmlTaxonRelationsImport
|
349
|
return;
|
350
|
}
|
351
|
|
352
|
|
353
|
|
354
|
private void makeSpecimenCircumscription(TaxonBase name, Element elSpecimenCircumscription, ResultWrapper<Boolean> success){
|
355
|
if (elSpecimenCircumscription != null){
|
356
|
logger.warn("makeProviderLink not yet implemented");
|
357
|
success.setValue(false);
|
358
|
}
|
359
|
}
|
360
|
|
361
|
|
362
|
private void makeCharacterCircumscription(TaxonBase name, Element elCharacterCircumscription, ResultWrapper<Boolean> success){
|
363
|
if (elCharacterCircumscription != null){
|
364
|
logger.warn("makeProviderLink not yet implemented");
|
365
|
success.setValue(false);
|
366
|
}
|
367
|
}
|
368
|
|
369
|
private void makeProviderLink(TaxonBase name, Element elProviderLink, ResultWrapper<Boolean> success){
|
370
|
if (elProviderLink != null){
|
371
|
logger.warn("makeProviderLink not yet implemented");
|
372
|
success.setValue(false);
|
373
|
}
|
374
|
}
|
375
|
|
376
|
|
377
|
private void makeProviderSpecificData(TaxonBase name, Element elProviderSpecificData, ResultWrapper<Boolean> success){
|
378
|
if (elProviderSpecificData != null){
|
379
|
logger.warn("makeProviderLink not yet implemented");
|
380
|
success.setValue(false);
|
381
|
}
|
382
|
}
|
383
|
|
384
|
|
385
|
|
386
|
/* (non-Javadoc)
|
387
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
388
|
*/
|
389
|
protected boolean isIgnore(TcsXmlImportState state){
|
390
|
return ! state.getConfig().isDoTaxa();
|
391
|
}
|
392
|
|
393
|
|
394
|
}
|