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.sdd.in;
|
11
|
|
12
|
import java.io.File;
|
13
|
import java.net.MalformedURLException;
|
14
|
import java.net.URI;
|
15
|
import java.net.URL;
|
16
|
import java.text.SimpleDateFormat;
|
17
|
import java.util.ArrayList;
|
18
|
import java.util.Date;
|
19
|
import java.util.HashMap;
|
20
|
import java.util.HashSet;
|
21
|
import java.util.List;
|
22
|
import java.util.Map;
|
23
|
import java.util.Set;
|
24
|
|
25
|
import org.apache.commons.lang.StringUtils;
|
26
|
import org.apache.log4j.Logger;
|
27
|
import org.jdom.Element;
|
28
|
import org.jdom.Namespace;
|
29
|
import org.joda.time.DateTime;
|
30
|
import org.springframework.stereotype.Component;
|
31
|
import org.springframework.transaction.TransactionStatus;
|
32
|
|
33
|
import eu.etaxonomy.cdm.api.service.IDescriptionService;
|
34
|
import eu.etaxonomy.cdm.common.IProgressMonitor;
|
35
|
import eu.etaxonomy.cdm.common.media.ImageInfo;
|
36
|
import eu.etaxonomy.cdm.hibernate.HibernateProxyHelper;
|
37
|
import eu.etaxonomy.cdm.io.common.ICdmImport;
|
38
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
39
|
import eu.etaxonomy.cdm.io.common.XmlImportBase;
|
40
|
import eu.etaxonomy.cdm.io.sdd.SDDTransformer;
|
41
|
import eu.etaxonomy.cdm.model.agent.Person;
|
42
|
import eu.etaxonomy.cdm.model.agent.Team;
|
43
|
import eu.etaxonomy.cdm.model.common.Annotation;
|
44
|
import eu.etaxonomy.cdm.model.common.AnnotationType;
|
45
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
46
|
import eu.etaxonomy.cdm.model.common.DefinedTermBase;
|
47
|
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
|
48
|
import eu.etaxonomy.cdm.model.common.IdentifiableSource;
|
49
|
import eu.etaxonomy.cdm.model.common.Language;
|
50
|
import eu.etaxonomy.cdm.model.common.LanguageString;
|
51
|
import eu.etaxonomy.cdm.model.common.Marker;
|
52
|
import eu.etaxonomy.cdm.model.common.MarkerType;
|
53
|
import eu.etaxonomy.cdm.model.common.Representation;
|
54
|
import eu.etaxonomy.cdm.model.common.TermBase;
|
55
|
import eu.etaxonomy.cdm.model.common.TermVocabulary;
|
56
|
import eu.etaxonomy.cdm.model.common.VersionableEntity;
|
57
|
import eu.etaxonomy.cdm.model.description.CategoricalData;
|
58
|
import eu.etaxonomy.cdm.model.description.Feature;
|
59
|
import eu.etaxonomy.cdm.model.description.FeatureNode;
|
60
|
import eu.etaxonomy.cdm.model.description.FeatureTree;
|
61
|
import eu.etaxonomy.cdm.model.description.MeasurementUnit;
|
62
|
import eu.etaxonomy.cdm.model.description.Modifier;
|
63
|
import eu.etaxonomy.cdm.model.description.QuantitativeData;
|
64
|
import eu.etaxonomy.cdm.model.description.State;
|
65
|
import eu.etaxonomy.cdm.model.description.StateData;
|
66
|
import eu.etaxonomy.cdm.model.description.StatisticalMeasure;
|
67
|
import eu.etaxonomy.cdm.model.description.StatisticalMeasurementValue;
|
68
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
69
|
import eu.etaxonomy.cdm.model.description.TextData;
|
70
|
import eu.etaxonomy.cdm.model.description.WorkingSet;
|
71
|
import eu.etaxonomy.cdm.model.location.NamedArea;
|
72
|
import eu.etaxonomy.cdm.model.media.IdentifiableMediaEntity;
|
73
|
import eu.etaxonomy.cdm.model.media.ImageFile;
|
74
|
import eu.etaxonomy.cdm.model.media.Media;
|
75
|
import eu.etaxonomy.cdm.model.media.MediaRepresentation;
|
76
|
import eu.etaxonomy.cdm.model.media.MediaRepresentationPart;
|
77
|
import eu.etaxonomy.cdm.model.media.Rights;
|
78
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
79
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
80
|
import eu.etaxonomy.cdm.model.occurrence.Specimen;
|
81
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
82
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
83
|
import eu.etaxonomy.cdm.model.taxon.Classification;
|
84
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
85
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
86
|
import eu.etaxonomy.cdm.model.taxon.TaxonNode;
|
87
|
|
88
|
/**
|
89
|
* @author h.fradin
|
90
|
* @created 24.10.2008
|
91
|
* @version 1.0
|
92
|
*/
|
93
|
@Component("sddImport")
|
94
|
public class SDDImport extends XmlImportBase<SDDImportConfigurator, SDDImportState> implements ICdmImport<SDDImportConfigurator, SDDImportState> {
|
95
|
private static final Logger logger = Logger.getLogger(SDDImport.class);
|
96
|
|
97
|
private static int modCount = 1000;
|
98
|
|
99
|
private Map<String,Person> authors = new HashMap<String,Person>();
|
100
|
private Map<String,String> citations = new HashMap<String,String>();
|
101
|
private Map<String,String> defaultUnitPrefixes = new HashMap<String,String>();
|
102
|
private Map<String,Person> editors = new HashMap<String,Person>();
|
103
|
private Map<String,FeatureNode> featureNodes = new HashMap<String,FeatureNode>();
|
104
|
private Map<String,Feature> features = new HashMap<String,Feature>();
|
105
|
private Map<String,String> locations = new HashMap<String,String>();
|
106
|
private Map<String,List<CdmBase>> mediaObject_ListCdmBase = new HashMap<String,List<CdmBase>>();
|
107
|
private Map<String,String> mediaObject_Role = new HashMap<String,String>();
|
108
|
private Map<String,Reference> publications = new HashMap<String,Reference>();
|
109
|
private Map<String,State> states = new HashMap<String,State>();
|
110
|
private Map<String,TaxonDescription> taxonDescriptions = new HashMap<String,TaxonDescription>();
|
111
|
private Map<String,NonViralName> taxonNameBases = new HashMap<String,NonViralName>();
|
112
|
private Map<String,MeasurementUnit> units = new HashMap<String,MeasurementUnit>();
|
113
|
private Map<String,TaxonNode> taxonNodes = new HashMap<String,TaxonNode>();
|
114
|
private Map<String,NamedArea> namedAreas = new HashMap<String,NamedArea>();
|
115
|
private Map<String,Specimen> specimens = new HashMap<String,Specimen>();
|
116
|
private Map<String,Modifier> modifiers = new HashMap<String,Modifier>();
|
117
|
|
118
|
private Set<MarkerType> markerTypes = new HashSet<MarkerType>();
|
119
|
private Set<TermVocabulary> vocabularies = new HashSet<TermVocabulary>();
|
120
|
|
121
|
private Set<Feature> descriptiveConcepts = new HashSet<Feature>();
|
122
|
private Set<AnnotationType> annotationTypes = new HashSet<AnnotationType>();
|
123
|
// private Set<Feature> featureSet = new HashSet<Feature>();
|
124
|
private Set<Reference> sources = new HashSet<Reference>();
|
125
|
private Reference sec = ReferenceFactory.newDatabase();
|
126
|
private Reference sourceReference = null;
|
127
|
|
128
|
private Language datasetLanguage = null;
|
129
|
private WorkingSet workingSet = null;
|
130
|
|
131
|
private Namespace xmlNamespace = Namespace.getNamespace("xml","http://www.w3.org/XML/1998/namespace");
|
132
|
|
133
|
private String generatorName = "";
|
134
|
private String generatorVersion = "";
|
135
|
|
136
|
|
137
|
private Set<StatisticalMeasure> statisticalMeasures = new HashSet<StatisticalMeasure>();
|
138
|
private Set<VersionableEntity> featureData = new HashSet<VersionableEntity>();
|
139
|
private Set<FeatureTree> featureTrees = new HashSet<FeatureTree>();
|
140
|
private Set<Classification> classifications = new HashSet<Classification>();
|
141
|
|
142
|
private Rights copyright = null;
|
143
|
|
144
|
private int taxonNamesCount = 0;
|
145
|
|
146
|
public SDDImport(){
|
147
|
super();
|
148
|
}
|
149
|
|
150
|
@Override
|
151
|
public boolean doCheck(SDDImportState state){
|
152
|
boolean result = true;
|
153
|
logger.warn("No check implemented for SDD");
|
154
|
return result;
|
155
|
}
|
156
|
|
157
|
// @Override
|
158
|
// public boolean doInvoke(IImportConfigurator config, Map<String, MapWrapper<? extends CdmBase>> stores){
|
159
|
@Override
|
160
|
public void doInvoke(SDDImportState state){
|
161
|
|
162
|
TransactionStatus ts = startTransaction();
|
163
|
SDDImportConfigurator sddConfig = state.getConfig();
|
164
|
IProgressMonitor progressMonitor = sddConfig.getProgressMonitor();
|
165
|
|
166
|
logger.info("start Datasets ...");
|
167
|
|
168
|
// <Datasets xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://rs.tdwg.org/UBIF/2006/" xsi:schemaLocation="http://rs.tdwg.org/UBIF/2006/ ../SDD.xsd">
|
169
|
Element root = sddConfig.getSourceRoot();
|
170
|
Namespace sddNamespace = sddConfig.getSddNamespace();
|
171
|
|
172
|
logger.info("start TechnicalMetadata ...");
|
173
|
// <TechnicalMetadata created="2006-04-20T10:00:00">
|
174
|
importTechnicalMetadata(root, sddNamespace, sddConfig);
|
175
|
List<Element> elDatasets = root.getChildren("Dataset",sddNamespace);
|
176
|
int i = 0;
|
177
|
|
178
|
//for each Dataset
|
179
|
logger.info("start Dataset ...");
|
180
|
progressMonitor.beginTask("Importing SDD data", elDatasets.size());
|
181
|
for (Element elDataset : elDatasets){
|
182
|
importDataset(elDataset, sddNamespace, state);
|
183
|
// if ((++i % modCount) == 0){ logger.info("dataset(s) handled: " + i);}
|
184
|
// logger.info(i + " dataset(s) handled");
|
185
|
progressMonitor.worked(1);
|
186
|
}
|
187
|
commitTransaction(ts);
|
188
|
progressMonitor.done();
|
189
|
logger.info("End of transaction");
|
190
|
return;
|
191
|
}
|
192
|
|
193
|
/* (non-Javadoc)
|
194
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
195
|
*/
|
196
|
protected boolean isIgnore(SDDImportState state){
|
197
|
return false;
|
198
|
}
|
199
|
|
200
|
|
201
|
// associates the reference of a media object in SDD with a CdmBase Object
|
202
|
protected void associateImageWithCdmBase(String refMO, CdmBase cb){
|
203
|
if ((refMO != null) && (cb!=null)) {
|
204
|
if (! refMO.equals("")) {
|
205
|
if (! mediaObject_ListCdmBase.containsKey(refMO)) {
|
206
|
List<CdmBase> lcb = new ArrayList<CdmBase>();
|
207
|
lcb.add(cb);
|
208
|
mediaObject_ListCdmBase.put(refMO,lcb);
|
209
|
} else {
|
210
|
List<CdmBase> lcb = mediaObject_ListCdmBase.get(refMO);
|
211
|
lcb.add(cb);
|
212
|
mediaObject_ListCdmBase.put(refMO,lcb);
|
213
|
}
|
214
|
}
|
215
|
}
|
216
|
}
|
217
|
|
218
|
// imports information about the Dataset
|
219
|
protected void importDatasetRepresentation(Element parent, Namespace sddNamespace){
|
220
|
logger.info("start Representation ...");
|
221
|
/* <Representation>
|
222
|
<Label>The Genus Viola</Label>
|
223
|
<Detail>This is an example for a very simple SDD file, representing a single description with categorical, quantitative, and text character. Compare also the "Fragment*" examples, which contain more complex examples in the form of document fragments. Intended for version="SDD 1.1".</Detail>
|
224
|
</Representation>
|
225
|
*/
|
226
|
|
227
|
|
228
|
|
229
|
Element elRepresentation = parent.getChild("Representation",sddNamespace);
|
230
|
String label = (String)ImportHelper.getXmlInputValue(elRepresentation, "Label",sddNamespace);
|
231
|
String detail = (String)ImportHelper.getXmlInputValue(elRepresentation, "Detail",sddNamespace);
|
232
|
|
233
|
//new
|
234
|
Representation representation = Representation.NewInstance(detail, label, null, datasetLanguage);
|
235
|
workingSet.addRepresentation(representation);
|
236
|
|
237
|
|
238
|
//old
|
239
|
// sec.setTitleCache(label, true);
|
240
|
//
|
241
|
// if (detail != null) {
|
242
|
// Annotation annotation = Annotation.NewInstance(detail, datasetLanguage);
|
243
|
// annotation.setAnnotationType(AnnotationType.EDITORIAL());
|
244
|
// sec.addAnnotation(annotation);
|
245
|
// }
|
246
|
|
247
|
|
248
|
List<Element> listMediaObjects = elRepresentation.getChildren("MediaObject",sddNamespace);
|
249
|
|
250
|
for (Element elMediaObject : listMediaObjects) {
|
251
|
String ref = null;
|
252
|
String role = null;
|
253
|
if (elMediaObject != null) {
|
254
|
ref = elMediaObject.getAttributeValue("ref");
|
255
|
role = elMediaObject.getAttributeValue("role");
|
256
|
}
|
257
|
if (ref != null) {
|
258
|
if (!ref.equals("")) {
|
259
|
this.associateImageWithCdmBase(ref,sourceReference);
|
260
|
this.associateImageWithCdmBase(ref,sec);
|
261
|
mediaObject_Role.put(ref,role);
|
262
|
}
|
263
|
}
|
264
|
}
|
265
|
}
|
266
|
|
267
|
// imports the representation (label, detail, lang) of a particular SDD element
|
268
|
protected void importRepresentation(Element parent, Namespace sddNamespace, VersionableEntity ve, String id, SDDImportState state){
|
269
|
Element elRepresentation = parent.getChild("Representation",sddNamespace);
|
270
|
|
271
|
Map<Language,List<String>> langLabDet = new HashMap<Language,List<String>>();
|
272
|
|
273
|
handleRepresentationLabels(sddNamespace, elRepresentation, langLabDet);
|
274
|
handleRepresentationDetails(sddNamespace, elRepresentation, langLabDet);
|
275
|
|
276
|
if (ve instanceof TermBase) {
|
277
|
makeRepresentationForTerms((TermBase)ve, langLabDet);
|
278
|
}else if (ve instanceof Media) {
|
279
|
makeRepresentationForMedia((Media)ve, langLabDet);
|
280
|
}else if (ve instanceof IdentifiableEntity<?>) {
|
281
|
IdentifiableEntity<?> ie = (IdentifiableEntity<?>)ve;
|
282
|
makeRepresentationForIdentifiableEntity(sddNamespace, ie, elRepresentation, langLabDet);
|
283
|
if (ve instanceof IdentifiableMediaEntity<?>){
|
284
|
makeRepresentationForIdentifiableMediaEntity(parent, sddNamespace, (IdentifiableMediaEntity<?>)ve);
|
285
|
}
|
286
|
}
|
287
|
|
288
|
makeRepresentationMediaObjects(sddNamespace, ve, elRepresentation);//FIXME
|
289
|
|
290
|
}
|
291
|
|
292
|
|
293
|
/**
|
294
|
* Handles the "Detail" children of representations. Adds the result to the langLabDet.
|
295
|
* @param sddNamespace
|
296
|
* @param elRepresentation
|
297
|
* @param langLabDet
|
298
|
*/
|
299
|
private void handleRepresentationDetails(Namespace sddNamespace,
|
300
|
Element elRepresentation, Map<Language, List<String>> langLabDet) {
|
301
|
List<Element> listDetails = elRepresentation.getChildren("Detail",sddNamespace);
|
302
|
for (Element elDetail : listDetails){
|
303
|
Language language = getLanguage(elDetail);
|
304
|
String role = elDetail.getAttributeValue("role");
|
305
|
String detail = elDetail.getText();
|
306
|
List<String> labDet = langLabDet.get(language);
|
307
|
labDet.add(detail);
|
308
|
labDet.add(role);
|
309
|
langLabDet.put(language, labDet);
|
310
|
}
|
311
|
}
|
312
|
|
313
|
/**
|
314
|
* Handles the "Label" children of representations. Adds the result to the langLabDet.
|
315
|
* @param sddNamespace
|
316
|
* @param elRepresentation
|
317
|
* @param langLabDet
|
318
|
*/
|
319
|
private void handleRepresentationLabels(Namespace sddNamespace,
|
320
|
Element elRepresentation, Map<Language, List<String>> langLabDet) {
|
321
|
// <Label xml:lang="la">Viola hederacea Labill.</Label>
|
322
|
List<Element> listLabels = elRepresentation.getChildren("Label",sddNamespace);
|
323
|
for (Element elLabel : listLabels){
|
324
|
Language language = getLanguage(elLabel);
|
325
|
String label = elLabel.getText();
|
326
|
List<String> labDet = new ArrayList<String>(3);
|
327
|
labDet.add(label);
|
328
|
langLabDet.put(language, labDet);
|
329
|
}
|
330
|
}
|
331
|
|
332
|
/**
|
333
|
*
|
334
|
* @param ve
|
335
|
* @param langLabDet
|
336
|
*/
|
337
|
private void makeRepresentationForMedia(Media m, Map<Language, List<String>> langLabDet) {
|
338
|
for (Language lang : langLabDet.keySet()){
|
339
|
List<String> labDet = langLabDet.get(lang);
|
340
|
if (labDet.get(0) != null){
|
341
|
m.putTitle(LanguageString.NewInstance(labDet.get(0), lang));
|
342
|
}
|
343
|
if (labDet.size()>1) {
|
344
|
m.addDescription(labDet.get(1), lang);
|
345
|
}
|
346
|
}
|
347
|
}
|
348
|
|
349
|
/**
|
350
|
* Handles representations for terms. Adds one representation per language in langLabDet.
|
351
|
*
|
352
|
* @param ve
|
353
|
* @param langLabDet
|
354
|
*/
|
355
|
private void makeRepresentationForTerms(TermBase tb, Map<Language, List<String>> langLabDet) {
|
356
|
for (Language lang : langLabDet.keySet()){
|
357
|
List<String> labDet = langLabDet.get(lang);
|
358
|
if (labDet.size()>0){
|
359
|
if (labDet.size()>1) {
|
360
|
tb.addRepresentation(Representation.NewInstance(labDet.get(1), labDet.get(0), labDet.get(0), lang));
|
361
|
} else {
|
362
|
tb.addRepresentation(Representation.NewInstance(labDet.get(0), labDet.get(0), labDet.get(0), lang));
|
363
|
}
|
364
|
}
|
365
|
}
|
366
|
}
|
367
|
|
368
|
|
369
|
/**
|
370
|
* Handles the "MediaObject" children of representations.
|
371
|
* @param sddNamespace
|
372
|
* @param ve
|
373
|
* @param elRepresentation
|
374
|
*/
|
375
|
private void makeRepresentationMediaObjects(Namespace sddNamespace, VersionableEntity ve, Element elRepresentation) {
|
376
|
List <Element> listMediaObjects = elRepresentation.getChildren("MediaObject", sddNamespace);
|
377
|
for (Element elMediaObject : listMediaObjects) {
|
378
|
String ref = null;
|
379
|
//TODO
|
380
|
String role = null;
|
381
|
if (elMediaObject != null) {
|
382
|
ref = elMediaObject.getAttributeValue("ref");
|
383
|
role = elMediaObject.getAttributeValue("role");
|
384
|
}
|
385
|
if (StringUtils.isNotBlank(ref)) {
|
386
|
if (ve instanceof TaxonDescription) {
|
387
|
TaxonDescription td = (TaxonDescription) ve;
|
388
|
if (td.getDescriptionSources().size() > 0) {
|
389
|
this.associateImageWithCdmBase(ref,(Reference) td.getDescriptionSources().toArray()[0]);
|
390
|
} else {
|
391
|
Reference descriptionSource = ReferenceFactory.newGeneric();
|
392
|
sources.add(descriptionSource);
|
393
|
td.addDescriptionSource(descriptionSource);
|
394
|
this.associateImageWithCdmBase(ref,descriptionSource);
|
395
|
}
|
396
|
} else {
|
397
|
this.associateImageWithCdmBase(ref,ve);
|
398
|
}
|
399
|
}
|
400
|
}
|
401
|
}
|
402
|
|
403
|
/**
|
404
|
* Handles the "Links" element
|
405
|
* @param parent
|
406
|
* @param sddNamespace
|
407
|
* @param ve
|
408
|
*/
|
409
|
private void makeRepresentationForIdentifiableMediaEntity(Element parent,
|
410
|
Namespace sddNamespace, IdentifiableMediaEntity ime) {
|
411
|
Element elLinks = parent.getChild("Links",sddNamespace);
|
412
|
|
413
|
if (elLinks != null) {
|
414
|
|
415
|
// <Link rel="Alternate" href="http://www.diversitycampus.net/people/hagedorn"/>
|
416
|
List<Element> listLinks = elLinks.getChildren("Link", sddNamespace);
|
417
|
Media link = Media.NewInstance();
|
418
|
MediaRepresentation mr = MediaRepresentation.NewInstance();
|
419
|
int k = 0;
|
420
|
//for each Link
|
421
|
for (Element elLink : listLinks){
|
422
|
|
423
|
try {
|
424
|
//TODO
|
425
|
String rel = elLink.getAttributeValue("rel");
|
426
|
String href = elLink.getAttributeValue("href");
|
427
|
URI uri = new URI(href);
|
428
|
mr.addRepresentationPart(MediaRepresentationPart.NewInstance(uri, null));
|
429
|
link.addRepresentation(mr);
|
430
|
ime.addMedia(link);
|
431
|
|
432
|
} catch (Exception e) {
|
433
|
//FIXME
|
434
|
logger.warn("Import of Link " + k + " failed.");
|
435
|
}
|
436
|
|
437
|
if ((++k % modCount) == 0){ logger.info("Links handled: " + k);}
|
438
|
|
439
|
}
|
440
|
}
|
441
|
}
|
442
|
|
443
|
/**
|
444
|
* @param sddNamespace
|
445
|
* @param ve
|
446
|
* @param elRepresentation
|
447
|
* @param langLabDet
|
448
|
* @return
|
449
|
*/
|
450
|
private void makeRepresentationForIdentifiableEntity(Namespace sddNamespace, IdentifiableEntity<?> ie,
|
451
|
Element elRepresentation, Map<Language, List<String>> langLabDet) {
|
452
|
List<String> labDet = null;
|
453
|
|
454
|
if (ie instanceof TaxonNameBase) {
|
455
|
if (langLabDet.keySet().contains(getTermService().getLanguageByIso("la"))) {
|
456
|
labDet = langLabDet.get(getTermService().getLanguageByIso("la"));
|
457
|
} else if (langLabDet.keySet().contains(datasetLanguage)) {
|
458
|
labDet = langLabDet.get(datasetLanguage);
|
459
|
logger.info("TaxonName " + (String)ImportHelper.getXmlInputValue(elRepresentation, "Label",sddNamespace) + " is not specified as a latin name.");
|
460
|
} else {
|
461
|
labDet = langLabDet.get(langLabDet.keySet().iterator().next());
|
462
|
logger.info("TaxonName " + (String)ImportHelper.getXmlInputValue(elRepresentation, "Label",sddNamespace) + " is not specified as a latin name.");
|
463
|
}
|
464
|
} else {
|
465
|
labDet = langLabDet.get(langLabDet.keySet().iterator().next());
|
466
|
}
|
467
|
|
468
|
//FIXME labDet is != null only for TaxonNameBase
|
469
|
ie.setTitleCache(labDet.get(0), true);
|
470
|
|
471
|
if (labDet.size()>1) {
|
472
|
Annotation annotation = null;
|
473
|
if (labDet.get(1) != null) {
|
474
|
if (labDet.get(2) != null) {
|
475
|
annotation = Annotation.NewInstance(labDet.get(2) + " - " + labDet.get(1), datasetLanguage);
|
476
|
} else {
|
477
|
annotation = Annotation.NewInstance(labDet.get(1), datasetLanguage);
|
478
|
}
|
479
|
}
|
480
|
ie.addAnnotation(annotation);
|
481
|
}
|
482
|
return;
|
483
|
}
|
484
|
|
485
|
/**
|
486
|
* @param elLabel
|
487
|
* @return
|
488
|
*/
|
489
|
private Language getLanguage(Element elLanguage) {
|
490
|
String lang = elLanguage.getAttributeValue("lang",xmlNamespace);
|
491
|
Language language = null;
|
492
|
if (StringUtils.isNotBlank(lang)) {
|
493
|
language = getTermService().getLanguageByIso(lang.substring(0, 2));
|
494
|
} else {
|
495
|
language = datasetLanguage;
|
496
|
}
|
497
|
return language;
|
498
|
}
|
499
|
|
500
|
|
501
|
// imports the representation (label, detail, lang) of a particular SDD element
|
502
|
protected void importTechnicalMetadata(Element root, Namespace sddNamespace, SDDImportConfigurator sddConfig){
|
503
|
Element elTechnicalMetadata = root.getChild("TechnicalMetadata", sddNamespace);
|
504
|
String nameCreated = elTechnicalMetadata.getAttributeValue("created");
|
505
|
sourceReference = sddConfig.getSourceReference();
|
506
|
|
507
|
if (nameCreated != null) {
|
508
|
if (!nameCreated.equals("")) {
|
509
|
int year = Integer.parseInt(nameCreated.substring(0,4));
|
510
|
int monthOfYear = Integer.parseInt(nameCreated.substring(5,7));
|
511
|
int dayOfMonth = Integer.parseInt(nameCreated.substring(8,10));
|
512
|
int hourOfDay = Integer.parseInt(nameCreated.substring(11,13));
|
513
|
int minuteOfHour = Integer.parseInt(nameCreated.substring(14,16));
|
514
|
int secondOfMinute = Integer.parseInt(nameCreated.substring(17,19));
|
515
|
DateTime created = new DateTime(year,monthOfYear,dayOfMonth,hourOfDay,minuteOfHour,secondOfMinute,0);
|
516
|
sourceReference.setCreated(created);
|
517
|
sec.setCreated(created);
|
518
|
}
|
519
|
}
|
520
|
|
521
|
// <Generator name="n/a, handcrafted instance document" version="n/a"/>
|
522
|
Element elGenerator = elTechnicalMetadata.getChild("Generator", sddNamespace);
|
523
|
generatorName = elGenerator.getAttributeValue("name");
|
524
|
generatorVersion = elGenerator.getAttributeValue("version");
|
525
|
|
526
|
sec.addAnnotation(Annotation.NewDefaultLanguageInstance(generatorName + " - " + generatorVersion));
|
527
|
sourceReference.addAnnotation(Annotation.NewDefaultLanguageInstance(generatorName + " - " + generatorVersion));
|
528
|
|
529
|
}
|
530
|
|
531
|
// imports the complete dataset information
|
532
|
protected void importDataset(Element elDataset, Namespace sddNamespace, SDDImportState state){ // <Dataset xml:lang="en-us">
|
533
|
|
534
|
workingSet = WorkingSet.NewInstance();
|
535
|
importDatasetLanguage(elDataset,state);
|
536
|
importDatasetRepresentation(elDataset, sddNamespace);
|
537
|
importRevisionData(elDataset, sddNamespace);
|
538
|
importIPRStatements(elDataset, sddNamespace, state);
|
539
|
importTaxonNames(elDataset, sddNamespace, state);
|
540
|
|
541
|
importDescriptiveConcepts(elDataset, sddNamespace, state);
|
542
|
importCharacters(elDataset, sddNamespace, state);
|
543
|
importCharacterTrees(elDataset, sddNamespace, state);
|
544
|
|
545
|
MarkerType editorMarkerType = getMarkerType(state, SDDTransformer.uuidMarkerEditor, "editor", "Editor", "edt");
|
546
|
MarkerType geographicAreaMarkerType = getMarkerType(state, SDDTransformer.uuidMarkerSDDGeographicArea, "SDDGeographicArea", "SDDGeographicArea", "ga");
|
547
|
MarkerType descriptiveConceptMarkerType = getMarkerType(state, SDDTransformer.uuidMarkerDescriptiveConcept, "DescriptiveConcept", "Descriptive Concept", "DC");
|
548
|
markerTypes.add(editorMarkerType);
|
549
|
markerTypes.add(geographicAreaMarkerType);
|
550
|
markerTypes.add(descriptiveConceptMarkerType);
|
551
|
|
552
|
//saving of all imported data into the CDM db
|
553
|
saveVocabularies();
|
554
|
saveFeatures();
|
555
|
saveModifiers();
|
556
|
saveStates();
|
557
|
saveMarkerType();
|
558
|
saveAreas(geographicAreaMarkerType);
|
559
|
saveUnits();
|
560
|
saveStatisticalMeasure();
|
561
|
saveAnnotationType();
|
562
|
|
563
|
importCodedDescriptions(elDataset, sddNamespace, state);
|
564
|
importAgents(elDataset, sddNamespace, state);
|
565
|
importPublications(elDataset, sddNamespace, state);
|
566
|
importMediaObjects(elDataset, sddNamespace, state);
|
567
|
importTaxonHierarchies(elDataset, sddNamespace, state);
|
568
|
importGeographicAreas(elDataset, sddNamespace, state);
|
569
|
importSpecimens(elDataset,sddNamespace, state);
|
570
|
|
571
|
|
572
|
if ((authors != null)||(editors != null)) {
|
573
|
Team team = Team.NewInstance();
|
574
|
if (authors != null) {
|
575
|
for (Person author : authors.values()){
|
576
|
team.addTeamMember(author);
|
577
|
}
|
578
|
}
|
579
|
if (editors != null) {
|
580
|
Marker marker = Marker.NewInstance();
|
581
|
marker.setMarkerType(editorMarkerType);
|
582
|
for (Person editor : editors.values()){
|
583
|
Person edit = editor;
|
584
|
edit.addMarker(marker);
|
585
|
team.addTeamMember(edit);
|
586
|
}
|
587
|
}
|
588
|
sec.setAuthorTeam(team);
|
589
|
sourceReference.setAuthorTeam(team);
|
590
|
}
|
591
|
|
592
|
if (copyright != null) {
|
593
|
sourceReference.addRights(copyright);
|
594
|
sec.addRights(copyright);
|
595
|
}
|
596
|
|
597
|
// Returns a CdmApplicationController created by the values of this configuration.
|
598
|
IDescriptionService descriptionService = getDescriptionService();
|
599
|
|
600
|
for (TaxonDescription taxonDescription : taxonDescriptions.values()){
|
601
|
// Persists a Description
|
602
|
descriptionService.save(taxonDescription);
|
603
|
}
|
604
|
|
605
|
for (String ref : taxonDescriptions.keySet()){
|
606
|
TaxonDescription td = taxonDescriptions.get(ref);
|
607
|
if (citations.containsKey(ref)) {
|
608
|
Reference publication = publications.get(citations.get(ref));
|
609
|
if (locations.containsKey(ref)) {
|
610
|
Annotation location = Annotation.NewInstance(locations.get(ref), datasetLanguage);
|
611
|
AnnotationType annotationType = AnnotationType.NewInstance("", "location", "");
|
612
|
annotationTypes.add(annotationType);
|
613
|
location.setAnnotationType(annotationType);
|
614
|
(publication).addAnnotation(location);
|
615
|
}
|
616
|
td.addDescriptionSource(publication);
|
617
|
}
|
618
|
}
|
619
|
logger.info("end makeTaxonDescriptions ...");
|
620
|
|
621
|
if (descriptiveConcepts != null) {
|
622
|
for (Feature feature : descriptiveConcepts) {
|
623
|
Marker marker = Marker.NewInstance();
|
624
|
marker.setMarkerType(descriptiveConceptMarkerType);
|
625
|
feature.addMarker(marker);
|
626
|
}
|
627
|
}
|
628
|
saveFeatures();
|
629
|
|
630
|
for (Reference publication : publications.values()){
|
631
|
getReferenceService().save(publication);
|
632
|
}
|
633
|
|
634
|
for (Reference source : sources){
|
635
|
getReferenceService().save(source);
|
636
|
}
|
637
|
|
638
|
for (FeatureTree featureTree : featureTrees) {
|
639
|
getFeatureTreeService().save(featureTree);
|
640
|
}
|
641
|
getWorkingSetService().save(workingSet);
|
642
|
for (Classification classification : classifications) {
|
643
|
getClassificationService().save(classification);
|
644
|
}
|
645
|
for (Specimen specimen : specimens.values()) {
|
646
|
getOccurrenceService().save(specimen);
|
647
|
}
|
648
|
logger.info("end of persistence ...");
|
649
|
|
650
|
return;
|
651
|
}
|
652
|
|
653
|
/**
|
654
|
*
|
655
|
*/
|
656
|
private void saveVocabularies() {
|
657
|
for (TermVocabulary vocabulary : vocabularies ){
|
658
|
getVocabularyService().save(vocabulary);
|
659
|
}
|
660
|
|
661
|
}
|
662
|
|
663
|
private void saveAnnotationType() {
|
664
|
for (AnnotationType annotationType: annotationTypes){
|
665
|
getTermService().save(annotationType);
|
666
|
}
|
667
|
}
|
668
|
|
669
|
private void saveStatisticalMeasure() {
|
670
|
for (StatisticalMeasure sm : statisticalMeasures){
|
671
|
getTermService().save(sm);
|
672
|
}
|
673
|
}
|
674
|
|
675
|
private void saveUnits() {
|
676
|
if (units != null) {
|
677
|
for (MeasurementUnit unit : units.values()){
|
678
|
if (unit != null) {
|
679
|
getTermService().save(unit);
|
680
|
}
|
681
|
}
|
682
|
}
|
683
|
}
|
684
|
|
685
|
private void saveAreas(MarkerType geographicAreaMarkerType) {
|
686
|
for (NamedArea area : namedAreas.values() ){
|
687
|
Marker marker = Marker.NewInstance();
|
688
|
marker.setMarkerType(geographicAreaMarkerType);
|
689
|
area.addMarker(marker);
|
690
|
getTermService().save(area);
|
691
|
}
|
692
|
}
|
693
|
|
694
|
private void saveStates() {
|
695
|
for (State state : states.values() ){
|
696
|
getTermService().save(state);
|
697
|
}
|
698
|
}
|
699
|
|
700
|
private void saveMarkerType() {
|
701
|
for (MarkerType markerType : markerTypes){
|
702
|
getTermService().save(markerType);
|
703
|
}
|
704
|
}
|
705
|
|
706
|
private void saveModifiers() {
|
707
|
for (Modifier modifier : modifiers.values() ){
|
708
|
getTermService().save(modifier);
|
709
|
}
|
710
|
}
|
711
|
|
712
|
private void saveFeatures() {
|
713
|
for (Feature feature : features.values() ){
|
714
|
getTermService().save(feature);
|
715
|
}
|
716
|
}
|
717
|
|
718
|
// imports the default language of the dataset
|
719
|
protected void importDatasetLanguage(Element elDataset, SDDImportState state){
|
720
|
String nameLang = elDataset.getAttributeValue("lang",xmlNamespace);
|
721
|
|
722
|
if (StringUtils.isNotBlank(nameLang)) {
|
723
|
String iso = nameLang.substring(0, 2);
|
724
|
datasetLanguage = getTermService().getLanguageByIso(iso);
|
725
|
} else {
|
726
|
datasetLanguage = Language.DEFAULT();
|
727
|
}
|
728
|
if (datasetLanguage == null) {
|
729
|
datasetLanguage = Language.DEFAULT();
|
730
|
}
|
731
|
}
|
732
|
|
733
|
// imports the specimens
|
734
|
protected void importSpecimens(Element elDataset, Namespace sddNamespace, SDDImportState cdmState) {
|
735
|
logger.info("start Specimens ...");
|
736
|
/* <Specimens>
|
737
|
<Specimen id="sp1">
|
738
|
<Representation>
|
739
|
<Label>TJM45337</Label>
|
740
|
</Representation>
|
741
|
</Specimen>
|
742
|
</Specimens>
|
743
|
*/
|
744
|
Element elSpecimens = elDataset.getChild("Specimens",sddNamespace);
|
745
|
if (elSpecimens != null){
|
746
|
List<Element> listSpecimens = elSpecimens.getChildren("Specimen", sddNamespace);
|
747
|
int j = 0;
|
748
|
for (Element elSpecimen : listSpecimens) {
|
749
|
String id = elSpecimen.getAttributeValue("id");
|
750
|
Specimen specimen = null;
|
751
|
if (!id.equals("")) {
|
752
|
specimen = Specimen.NewInstance();
|
753
|
specimens.put(id,specimen);
|
754
|
importRepresentation(elSpecimen, sddNamespace, specimen, id, cdmState);
|
755
|
}
|
756
|
}
|
757
|
|
758
|
}
|
759
|
}
|
760
|
|
761
|
// imports the revision data associated with the Dataset (authors, modifications)
|
762
|
protected void importRevisionData(Element elDataset, Namespace sddNamespace){
|
763
|
// <RevisionData>
|
764
|
logger.info("start RevisionData ...");
|
765
|
Element elRevisionData = elDataset.getChild("RevisionData",sddNamespace);
|
766
|
if (elRevisionData != null){
|
767
|
// <Creators>
|
768
|
Element elCreators = elRevisionData.getChild("Creators",sddNamespace);
|
769
|
|
770
|
// <Agent role="aut" ref="a1"/>
|
771
|
List<Element> listAgents = elCreators.getChildren("Agent", sddNamespace);
|
772
|
|
773
|
int j = 0;
|
774
|
//for each Agent
|
775
|
for (Element elAgent : listAgents){
|
776
|
|
777
|
String role = elAgent.getAttributeValue("role");
|
778
|
String ref = elAgent.getAttributeValue("ref");
|
779
|
if (role.equals("aut")) {
|
780
|
if(!ref.equals("")) {
|
781
|
authors.put(ref, null);
|
782
|
}
|
783
|
}
|
784
|
if (role.equals("edt")) {
|
785
|
if(!ref.equals("")) {
|
786
|
editors.put(ref, null);
|
787
|
}
|
788
|
}
|
789
|
if ((++j % modCount) == 0){ logger.info("Agents handled: " + j);}
|
790
|
|
791
|
}
|
792
|
|
793
|
// <DateModified>2006-04-08T00:00:00</DateModified>
|
794
|
String stringDateModified = (String)ImportHelper.getXmlInputValue(elRevisionData, "DateModified",sddNamespace);
|
795
|
|
796
|
if (stringDateModified != null) {
|
797
|
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd'T'hh:mm:ss");
|
798
|
Date d = null;
|
799
|
try {
|
800
|
d = sdf.parse(stringDateModified);
|
801
|
} catch(Exception e) {
|
802
|
System.err.println("Exception :");
|
803
|
e.printStackTrace();
|
804
|
}
|
805
|
|
806
|
DateTime updated = null;
|
807
|
if (d != null) {
|
808
|
updated = new DateTime(d);
|
809
|
sourceReference.setUpdated(updated);
|
810
|
sec.setUpdated(updated);
|
811
|
}
|
812
|
}
|
813
|
}
|
814
|
}
|
815
|
|
816
|
// imports ipr statements associated with a dataset
|
817
|
protected void importIPRStatements(Element elDataset, Namespace sddNamespace, SDDImportState state){
|
818
|
// <IPRStatements>
|
819
|
logger.info("start IPRStatements ...");
|
820
|
Element elIPRStatements = elDataset.getChild("IPRStatements",sddNamespace);
|
821
|
// <IPRStatement role="Copyright">
|
822
|
if (elIPRStatements != null) {
|
823
|
List<Element> listIPRStatements = elIPRStatements.getChildren("IPRStatement", sddNamespace);
|
824
|
int j = 0;
|
825
|
//for each IPRStatement
|
826
|
|
827
|
for (Element elIPRStatement : listIPRStatements){
|
828
|
|
829
|
String role = elIPRStatement.getAttributeValue("role");
|
830
|
// <Label xml:lang="en-au">(c) 2003-2006 Centre for Occasional Botany.</Label>
|
831
|
Element elLabel = elIPRStatement.getChild("Label",sddNamespace);
|
832
|
String lang = "";
|
833
|
if (elLabel != null) {
|
834
|
lang = elLabel.getAttributeValue("lang",xmlNamespace);
|
835
|
}
|
836
|
String label = (String)ImportHelper.getXmlInputValue(elIPRStatement, "Label",sddNamespace);
|
837
|
|
838
|
if (role.equals("Copyright")) {
|
839
|
Language iprLanguage = null;
|
840
|
if (lang != null) {
|
841
|
if (!lang.equals("")) {
|
842
|
iprLanguage = getTermService().getLanguageByIso(lang.substring(0, 2));
|
843
|
} else {
|
844
|
iprLanguage = datasetLanguage;
|
845
|
}
|
846
|
}
|
847
|
if (iprLanguage == null) {
|
848
|
iprLanguage = datasetLanguage;
|
849
|
}
|
850
|
copyright = Rights.NewInstance(label, iprLanguage);
|
851
|
}
|
852
|
|
853
|
if (copyright != null) {
|
854
|
sourceReference.addRights(copyright);
|
855
|
sec.addRights(copyright);
|
856
|
}
|
857
|
|
858
|
if ((++j % modCount) == 0){ logger.info("IPRStatements handled: " + j);}
|
859
|
|
860
|
}
|
861
|
}
|
862
|
}
|
863
|
|
864
|
// imports the taxon names
|
865
|
protected void importTaxonNames(Element elDataset, Namespace sddNamespace, SDDImportState state){
|
866
|
// <TaxonNames>
|
867
|
logger.info("start TaxonNames ...");
|
868
|
Element elTaxonNames = elDataset.getChild("TaxonNames",sddNamespace);
|
869
|
// <TaxonName id="t1" uri="urn:lsid:authority:namespace:my-own-id">
|
870
|
if (elTaxonNames != null) {
|
871
|
List<Element> listTaxonNames = elTaxonNames.getChildren("TaxonName", sddNamespace);
|
872
|
int j = 0;
|
873
|
//for each TaxonName
|
874
|
for (Element elTaxonName : listTaxonNames){
|
875
|
|
876
|
String id = elTaxonName.getAttributeValue("id");
|
877
|
String uri = elTaxonName.getAttributeValue("uri");
|
878
|
|
879
|
NonViralName tnb = null;
|
880
|
if (!id.equals("")) {
|
881
|
tnb = NonViralName.NewInstance(null);
|
882
|
IdentifiableSource source = null;
|
883
|
if (uri != null) {
|
884
|
if (!uri.equals("")) {
|
885
|
source = IdentifiableSource.NewInstance(id, "TaxonName", ReferenceFactory.newGeneric(), uri);
|
886
|
}
|
887
|
} else {
|
888
|
source = IdentifiableSource.NewInstance(id, "TaxonName");
|
889
|
}
|
890
|
tnb.addSource(source);
|
891
|
taxonNameBases.put(id,tnb);
|
892
|
}
|
893
|
|
894
|
// <Representation>
|
895
|
// <Label xml:lang="la">Viola hederacea Labill.</Label>
|
896
|
importRepresentation(elTaxonName, sddNamespace, tnb, id, state);
|
897
|
|
898
|
if ((++j % modCount) == 0){ logger.info("TaxonNames handled: " + j);}
|
899
|
|
900
|
}
|
901
|
}
|
902
|
}
|
903
|
|
904
|
// imports the characters (categorical, quantitative and text ; sequence characters not supported) which correspond to CDM Features
|
905
|
protected void importCharacters(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
906
|
// <Characters>
|
907
|
logger.info("start Characters ...");
|
908
|
Element elCharacters = elDataset.getChild("Characters", sddNamespace);
|
909
|
|
910
|
// <CategoricalCharacter id="c1">
|
911
|
if (elCharacters != null) {
|
912
|
handleCategoricalData(sddNamespace, cdmState, elCharacters);
|
913
|
handleQuantitativeData(sddNamespace, cdmState, elCharacters);
|
914
|
handleTextCharacters(sddNamespace, cdmState, elCharacters);
|
915
|
}
|
916
|
|
917
|
/*for (Iterator<Feature> f = features.values().iterator() ; f.hasNext() ;){
|
918
|
featureSet.add(f.next()); //XIM Why this line ?
|
919
|
}*/
|
920
|
|
921
|
return;
|
922
|
|
923
|
}
|
924
|
|
925
|
/**
|
926
|
* @param sddNamespace
|
927
|
* @param sddConfig
|
928
|
* @param elCharacters
|
929
|
* @return
|
930
|
*/
|
931
|
private void handleCategoricalData(Namespace sddNamespace, SDDImportState cdmState, Element elCharacters) {
|
932
|
List<Element> elCategoricalCharacters = elCharacters.getChildren("CategoricalCharacter", sddNamespace);
|
933
|
int j = 0;
|
934
|
for (Element elCategoricalCharacter : elCategoricalCharacters){
|
935
|
try {
|
936
|
|
937
|
String idCC = elCategoricalCharacter.getAttributeValue("id");
|
938
|
Feature categoricalCharacter = Feature.NewInstance();
|
939
|
categoricalCharacter.setKindOf(Feature.DESCRIPTION());
|
940
|
importRepresentation(elCategoricalCharacter, sddNamespace, categoricalCharacter, idCC, cdmState);
|
941
|
categoricalCharacter.setSupportsCategoricalData(true);
|
942
|
|
943
|
// <States>
|
944
|
Element elStates = elCategoricalCharacter.getChild("States",sddNamespace);
|
945
|
|
946
|
// <StateDefinition id="s1">
|
947
|
List<Element> elStateDefinitions = elStates.getChildren("StateDefinition",sddNamespace);
|
948
|
TermVocabulary<State> termVocabularyState = TermVocabulary.NewInstance(null, null, null, null);
|
949
|
|
950
|
vocabularies.add(termVocabularyState);
|
951
|
|
952
|
int k = 0;
|
953
|
//for each StateDefinition
|
954
|
for (Element elStateDefinition : elStateDefinitions){
|
955
|
|
956
|
if ((++k % modCount) == 0){ logger.info("StateDefinitions handled: " + (k-1));}
|
957
|
|
958
|
String idS = elStateDefinition.getAttributeValue("id");
|
959
|
State state = states.get(idS);
|
960
|
if (state == null){
|
961
|
state = State.NewInstance();
|
962
|
}else{
|
963
|
logger.debug("State duplicate found");
|
964
|
}
|
965
|
importRepresentation(elStateDefinition, sddNamespace, state, idS, cdmState);
|
966
|
|
967
|
termVocabularyState.addTerm(state);
|
968
|
states.put(idS,state);
|
969
|
}
|
970
|
categoricalCharacter.addSupportedCategoricalEnumeration(termVocabularyState);
|
971
|
features.put(idCC, categoricalCharacter);
|
972
|
|
973
|
} catch (Exception e) {
|
974
|
logger.warn("Import of CategoricalCharacter " + j + " failed.");
|
975
|
cdmState.setUnsuccessfull();
|
976
|
}
|
977
|
|
978
|
if ((++j % modCount) == 0){ logger.info("CategoricalCharacters handled: " + j);}
|
979
|
|
980
|
}
|
981
|
return;
|
982
|
}
|
983
|
|
984
|
/**
|
985
|
* @param sddNamespace
|
986
|
* @param sddConfig
|
987
|
* @param elCharacters
|
988
|
*/
|
989
|
private void handleQuantitativeData(Namespace sddNamespace, SDDImportState cdmState, Element elCharacters) {
|
990
|
int j;
|
991
|
// <QuantitativeCharacter id="c2">
|
992
|
List<Element> elQuantitativeCharacters = elCharacters.getChildren("QuantitativeCharacter", sddNamespace);
|
993
|
j = 0;
|
994
|
//for each QuantitativeCharacter
|
995
|
for (Element elQuantitativeCharacter : elQuantitativeCharacters){
|
996
|
|
997
|
try {
|
998
|
|
999
|
String idQC = elQuantitativeCharacter.getAttributeValue("id");
|
1000
|
|
1001
|
// <Representation>
|
1002
|
// <Label>Leaf length</Label>
|
1003
|
// </Representation>
|
1004
|
Feature quantitativeCharacter = Feature.NewInstance();
|
1005
|
quantitativeCharacter.setKindOf(Feature.DESCRIPTION());
|
1006
|
importRepresentation(elQuantitativeCharacter, sddNamespace, quantitativeCharacter, idQC, cdmState);
|
1007
|
|
1008
|
quantitativeCharacter.setSupportsQuantitativeData(true);
|
1009
|
|
1010
|
// <MeasurementUnit>
|
1011
|
// <Label role="Abbrev">m</Label>
|
1012
|
// </MeasurementUnit>
|
1013
|
Element elMeasurementUnit = elQuantitativeCharacter.getChild("MeasurementUnit",sddNamespace);
|
1014
|
String label = "";
|
1015
|
String role = "";
|
1016
|
if (elMeasurementUnit != null) {
|
1017
|
Element elLabel = elMeasurementUnit.getChild("Label",sddNamespace);
|
1018
|
role = elLabel.getAttributeValue("role");
|
1019
|
label = (String)ImportHelper.getXmlInputValue(elMeasurementUnit, "Label",sddNamespace);
|
1020
|
}
|
1021
|
|
1022
|
MeasurementUnit unit = null;
|
1023
|
if (!label.equals("")){
|
1024
|
if (role != null) {
|
1025
|
if (role.equals("Abbrev")){
|
1026
|
unit = MeasurementUnit.NewInstance(label,label,label);
|
1027
|
}
|
1028
|
} else {
|
1029
|
unit = MeasurementUnit.NewInstance(label,label,label);
|
1030
|
}
|
1031
|
}
|
1032
|
|
1033
|
if (unit != null) {
|
1034
|
units.put(idQC, unit);
|
1035
|
}
|
1036
|
|
1037
|
//<Default>
|
1038
|
// <MeasurementUnitPrefix>milli</MeasurementUnitPrefix>
|
1039
|
//</Default>
|
1040
|
Element elDefault = elQuantitativeCharacter.getChild("Default",sddNamespace);
|
1041
|
if (elDefault != null) {
|
1042
|
String measurementUnitPrefix = (String)ImportHelper.getXmlInputValue(elDefault, "MeasurementUnitPrefix",sddNamespace);
|
1043
|
if (! measurementUnitPrefix.equals("")){
|
1044
|
defaultUnitPrefixes.put(idQC, measurementUnitPrefix);
|
1045
|
}
|
1046
|
}
|
1047
|
|
1048
|
features.put(idQC, quantitativeCharacter);
|
1049
|
|
1050
|
} catch (Exception e) {
|
1051
|
//FIXME
|
1052
|
logger.warn("Import of QuantitativeCharacter " + j + " failed.");
|
1053
|
cdmState.setUnsuccessfull();
|
1054
|
}
|
1055
|
|
1056
|
if ((++j % modCount) == 0){ logger.info("QuantitativeCharacters handled: " + j);}
|
1057
|
|
1058
|
}
|
1059
|
return;
|
1060
|
}
|
1061
|
|
1062
|
private void handleTextCharacters(Namespace sddNamespace, SDDImportState cdmState, Element elCharacters) {
|
1063
|
int j;
|
1064
|
// <TextCharacter id="c3">
|
1065
|
List<Element> elTextCharacters = elCharacters.getChildren("TextCharacter", sddNamespace);
|
1066
|
j = 0;
|
1067
|
//for each TextCharacter
|
1068
|
for (Element elTextCharacter : elTextCharacters){
|
1069
|
|
1070
|
try {
|
1071
|
|
1072
|
String idTC = elTextCharacter.getAttributeValue("id");
|
1073
|
|
1074
|
// <Representation>
|
1075
|
// <Label xml:lang="en">Leaf features not covered by other characters</Label>
|
1076
|
// </Representation>
|
1077
|
Feature textCharacter = Feature.NewInstance();
|
1078
|
textCharacter.setKindOf(Feature.DESCRIPTION());
|
1079
|
importRepresentation(elTextCharacter, sddNamespace, textCharacter, idTC, cdmState);
|
1080
|
|
1081
|
textCharacter.setSupportsTextData(true);
|
1082
|
|
1083
|
features.put(idTC, textCharacter);
|
1084
|
|
1085
|
} catch (Exception e) {
|
1086
|
//FIXME
|
1087
|
logger.warn("Import of TextCharacter " + j + " failed.");
|
1088
|
cdmState.setUnsuccessfull();
|
1089
|
}
|
1090
|
|
1091
|
if ((++j % modCount) == 0){ logger.info("TextCharacters handled: " + j);}
|
1092
|
|
1093
|
}
|
1094
|
return;
|
1095
|
}
|
1096
|
|
1097
|
// imports the descriptions of taxa
|
1098
|
protected void importCodedDescriptions(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1099
|
|
1100
|
// <CodedDescriptions>
|
1101
|
logger.info("start CodedDescriptions ...");
|
1102
|
Element elCodedDescriptions = elDataset.getChild("CodedDescriptions",sddNamespace);
|
1103
|
|
1104
|
// <CodedDescription id="D101">
|
1105
|
if (elCodedDescriptions != null) {
|
1106
|
List<Element> listCodedDescriptions = elCodedDescriptions.getChildren("CodedDescription", sddNamespace);
|
1107
|
int j = 0;
|
1108
|
//for each CodedDescription
|
1109
|
for (Element elCodedDescription : listCodedDescriptions){
|
1110
|
handleCodedDescription(sddNamespace, cdmState, elCodedDescription, j);
|
1111
|
if ((++j % modCount) == 0){ logger.info("CodedDescriptions handled: " + j);}
|
1112
|
}
|
1113
|
}
|
1114
|
return;
|
1115
|
}
|
1116
|
|
1117
|
/**
|
1118
|
* @param sddNamespace
|
1119
|
* @param sddConfig
|
1120
|
* @param j
|
1121
|
* @param elCodedDescription
|
1122
|
* @return
|
1123
|
*/
|
1124
|
private void handleCodedDescription(Namespace sddNamespace, SDDImportState cdmState, Element elCodedDescription, int j) {
|
1125
|
try {
|
1126
|
|
1127
|
String idCD = elCodedDescription.getAttributeValue("id");
|
1128
|
|
1129
|
// <Representation>
|
1130
|
// <Label><i>Viola hederacea</i> Labill. as revised by R. Morris April 8, 2006</Label>
|
1131
|
// </Representation>
|
1132
|
TaxonDescription taxonDescription = TaxonDescription.NewInstance();
|
1133
|
if (!generatorName.isEmpty()){
|
1134
|
Annotation annotation = Annotation.NewInstance(generatorName, AnnotationType.TECHNICAL(),Language.DEFAULT());
|
1135
|
taxonDescription.addAnnotation(annotation);
|
1136
|
}
|
1137
|
importRepresentation(elCodedDescription, sddNamespace, taxonDescription, idCD, cdmState);
|
1138
|
|
1139
|
// <Scope>
|
1140
|
// <TaxonName ref="t1"/>
|
1141
|
// <Citation ref="p1" location="p. 30"/>
|
1142
|
// </Scope>
|
1143
|
Element elScope = elCodedDescription.getChild("Scope", sddNamespace);
|
1144
|
Taxon taxon;
|
1145
|
if (elScope != null) {
|
1146
|
taxon = handleCDScope(sddNamespace, cdmState, idCD, elScope);
|
1147
|
} else {//in case no taxon is linked to the description, a new one is created
|
1148
|
taxon = handleCDNoScope(sddNamespace, cdmState, elCodedDescription);
|
1149
|
}
|
1150
|
|
1151
|
// <SummaryData>
|
1152
|
Element elSummaryData = elCodedDescription.getChild("SummaryData",sddNamespace);
|
1153
|
if (elSummaryData != null) {
|
1154
|
handleSummaryCategoricalData(sddNamespace, taxonDescription, elSummaryData);
|
1155
|
handleSummaryQuantitativeData(sddNamespace, taxonDescription, elSummaryData);
|
1156
|
handleSummaryTextData(sddNamespace, taxonDescription, elSummaryData);
|
1157
|
}
|
1158
|
|
1159
|
if (taxon != null) {
|
1160
|
taxon.addDescription(taxonDescription);
|
1161
|
}
|
1162
|
//
|
1163
|
workingSet.addDescription(taxonDescription);
|
1164
|
|
1165
|
//OLD taxonDescription.setDescriptiveSystem(featureSet);
|
1166
|
|
1167
|
taxonDescriptions.put(idCD, taxonDescription);//FIXME
|
1168
|
|
1169
|
} catch (Exception e) {
|
1170
|
//FIXME
|
1171
|
logger.warn("Import of CodedDescription " + j + " failed.", e);
|
1172
|
cdmState.setUnsuccessfull();
|
1173
|
}
|
1174
|
return;
|
1175
|
}
|
1176
|
|
1177
|
/**
|
1178
|
* @param sddNamespace
|
1179
|
* @param sddConfig
|
1180
|
* @param elCodedDescription
|
1181
|
* @param taxon
|
1182
|
* @return
|
1183
|
*/
|
1184
|
private Taxon handleCDNoScope(Namespace sddNamespace, SDDImportState cdmState, Element elCodedDescription ) {
|
1185
|
Taxon taxon = null;
|
1186
|
NonViralName nonViralName = NonViralName.NewInstance(null);
|
1187
|
String id = new String("" + taxonNamesCount);
|
1188
|
IdentifiableSource source = IdentifiableSource.NewInstance(id, "TaxonName");
|
1189
|
importRepresentation(elCodedDescription, sddNamespace, nonViralName, id, cdmState);
|
1190
|
|
1191
|
if(cdmState.getConfig().isDoMatchTaxa()){
|
1192
|
taxon = getTaxonService().findBestMatchingTaxon(nonViralName.getTitleCache());
|
1193
|
}
|
1194
|
|
1195
|
if(taxon != null){
|
1196
|
nonViralName = HibernateProxyHelper.deproxy(taxon.getName(), NonViralName.class);
|
1197
|
// taxonNameBases.put(id ,tnb);
|
1198
|
// taxonNamesCount++;
|
1199
|
logger.info("using existing Taxon " + taxon.getTitleCache());
|
1200
|
} else {
|
1201
|
nonViralName.addSource(source);
|
1202
|
taxonNameBases.put(id ,nonViralName);
|
1203
|
taxonNamesCount++;
|
1204
|
logger.info("creating new Taxon from TaxonName " + nonViralName.getTitleCache());
|
1205
|
taxon = Taxon.NewInstance(nonViralName, sec);
|
1206
|
}
|
1207
|
return taxon;
|
1208
|
}
|
1209
|
|
1210
|
/**
|
1211
|
* @param sddNamespace
|
1212
|
* @param sddConfig
|
1213
|
* @param idCD
|
1214
|
* @param elScope
|
1215
|
* @param taxon
|
1216
|
* @return
|
1217
|
*/
|
1218
|
private Taxon handleCDScope(Namespace sddNamespace, SDDImportState cdmState,
|
1219
|
String idCD, Element elScope) {
|
1220
|
Taxon taxon = null;
|
1221
|
Element elTaxonName = elScope.getChild("TaxonName", sddNamespace);
|
1222
|
String ref = elTaxonName.getAttributeValue("ref");
|
1223
|
NonViralName nonViralName = taxonNameBases.get(ref);
|
1224
|
|
1225
|
if(cdmState.getConfig().isDoMatchTaxa()){
|
1226
|
taxon = getTaxonService().findBestMatchingTaxon(nonViralName.getTitleCache());
|
1227
|
}
|
1228
|
|
1229
|
if(taxon != null){
|
1230
|
logger.info("using existing Taxon" + taxon.getTitleCache());
|
1231
|
if(!nonViralName.getUuid().equals(taxon.getName().getUuid())){
|
1232
|
logger.warn("TaxonNameBase entity of existing taxon does not match Name in list -> replacing Name in list");
|
1233
|
nonViralName = HibernateProxyHelper.deproxy(taxon.getName(), NonViralName.class);
|
1234
|
}
|
1235
|
} else {
|
1236
|
logger.info("creating new Taxon from TaxonName '" + nonViralName.getTitleCache()+"'");
|
1237
|
taxon = Taxon.NewInstance(nonViralName, sec);
|
1238
|
}
|
1239
|
|
1240
|
//citation
|
1241
|
Element elCitation = elScope.getChild("Citation",sddNamespace);
|
1242
|
if (elCitation != null) {
|
1243
|
String refCitation = elCitation.getAttributeValue("ref");
|
1244
|
if (! refCitation.equals("")){
|
1245
|
citations.put(idCD, refCitation);
|
1246
|
}
|
1247
|
String location = elCitation.getAttributeValue("location");
|
1248
|
if (! location.equals("")){
|
1249
|
locations.put(idCD, location);
|
1250
|
}
|
1251
|
}
|
1252
|
return taxon;
|
1253
|
}
|
1254
|
|
1255
|
/**
|
1256
|
* @param sddNamespace
|
1257
|
* @param taxonDescription
|
1258
|
* @param elSummaryData
|
1259
|
*/
|
1260
|
private void handleSummaryTextData(Namespace sddNamespace,
|
1261
|
TaxonDescription taxonDescription, Element elSummaryData) {
|
1262
|
String ref;
|
1263
|
int k;
|
1264
|
// <TextChar ref="c3">
|
1265
|
List<Element> elTextChars = elSummaryData.getChildren("TextChar", sddNamespace);
|
1266
|
k = 0;
|
1267
|
//for each TextChar
|
1268
|
for (Element elTextChar : elTextChars){
|
1269
|
if ((++k % modCount) == 0){ logger.info("TextChar handled: " + (k-1));}
|
1270
|
ref = elTextChar.getAttributeValue("ref");
|
1271
|
Feature feature = features.get(ref);
|
1272
|
TextData textData = TextData.NewInstance();
|
1273
|
textData.setFeature(feature);
|
1274
|
|
1275
|
// <Content>Free form text</Content>
|
1276
|
String content = (String)ImportHelper.getXmlInputValue(elTextChar, "Content",sddNamespace);
|
1277
|
textData.putText(datasetLanguage, content);
|
1278
|
taxonDescription.addElement(textData);
|
1279
|
}
|
1280
|
}
|
1281
|
|
1282
|
/**
|
1283
|
* @param sddNamespace
|
1284
|
* @param taxonDescription
|
1285
|
* @param elSummaryData
|
1286
|
*/
|
1287
|
private void handleSummaryQuantitativeData(Namespace sddNamespace,
|
1288
|
TaxonDescription taxonDescription, Element elSummaryData) {
|
1289
|
String ref;
|
1290
|
int k;
|
1291
|
// <Quantitative ref="c2">
|
1292
|
List<Element> elQuantitatives = elSummaryData.getChildren("Quantitative", sddNamespace);
|
1293
|
k = 0;
|
1294
|
//for each Quantitative
|
1295
|
for (Element elQuantitative : elQuantitatives){
|
1296
|
if ((++k % modCount) == 0){ logger.warn("Quantitative handled: " + (k-1));}
|
1297
|
ref = elQuantitative.getAttributeValue("ref");
|
1298
|
Feature feature = features.get(ref);
|
1299
|
QuantitativeData quantitativeData = QuantitativeData.NewInstance();
|
1300
|
quantitativeData.setFeature(feature);
|
1301
|
|
1302
|
MeasurementUnit unit = units.get(ref);
|
1303
|
String prefix = defaultUnitPrefixes.get(ref);
|
1304
|
if (unit != null) {
|
1305
|
String u = unit.getLabel();
|
1306
|
if (prefix != null) {
|
1307
|
u = prefix + u;
|
1308
|
}
|
1309
|
unit.setLabel(u);
|
1310
|
quantitativeData.setUnit(unit);
|
1311
|
}
|
1312
|
|
1313
|
// <Measure type="Min" value="2.3"/>
|
1314
|
List<Element> elMeasures = elQuantitative.getChildren("Measure", sddNamespace);
|
1315
|
int l = 0;
|
1316
|
|
1317
|
//for each State
|
1318
|
for (Element elMeasure : elMeasures){
|
1319
|
if ((++l % modCount) == 0){ logger.info("States handled: " + (l-1));}
|
1320
|
String type = elMeasure.getAttributeValue("type");
|
1321
|
String value = elMeasure.getAttributeValue("value");
|
1322
|
if (value.contains(",")) {
|
1323
|
value = value.replace(',', '.');
|
1324
|
}
|
1325
|
Float v = Float.parseFloat(value);
|
1326
|
//Float v = new Float(0);
|
1327
|
StatisticalMeasure t = null;
|
1328
|
if (type.equals("Min")) {
|
1329
|
t = StatisticalMeasure.MIN();
|
1330
|
} else if (type.equals("Mean")) {
|
1331
|
t = StatisticalMeasure.AVERAGE();
|
1332
|
} else if (type.equals("Max")) {
|
1333
|
t = StatisticalMeasure.MAX();
|
1334
|
} else if (type.equals("SD")) {
|
1335
|
t = StatisticalMeasure.STANDARD_DEVIATION();
|
1336
|
} else if (type.equals("N")) {
|
1337
|
t = StatisticalMeasure.SAMPLE_SIZE();
|
1338
|
} else if (type.equals("UMethLower")) {
|
1339
|
t = StatisticalMeasure.TYPICAL_LOWER_BOUNDARY();
|
1340
|
} else if (type.equals("UMethUpper")) {
|
1341
|
t = StatisticalMeasure.TYPICAL_UPPER_BOUNDARY();
|
1342
|
} else if (type.equals("Var")) {
|
1343
|
t = StatisticalMeasure.VARIANCE();
|
1344
|
} else {
|
1345
|
t = StatisticalMeasure.NewInstance(type,type,type);
|
1346
|
statisticalMeasures.add(t);
|
1347
|
}
|
1348
|
|
1349
|
StatisticalMeasurementValue statisticalValue = StatisticalMeasurementValue.NewInstance();
|
1350
|
statisticalValue.setValue(v);
|
1351
|
statisticalValue.setType(t);
|
1352
|
quantitativeData.addStatisticalValue(statisticalValue);
|
1353
|
featureData.add(statisticalValue);
|
1354
|
}
|
1355
|
taxonDescription.addElement(quantitativeData);
|
1356
|
}
|
1357
|
}
|
1358
|
|
1359
|
/**
|
1360
|
* @param sddNamespace
|
1361
|
* @param taxonDescription
|
1362
|
* @param elSummaryData
|
1363
|
*/
|
1364
|
private void handleSummaryCategoricalData(Namespace sddNamespace,
|
1365
|
TaxonDescription taxonDescription, Element elSummaryData) {
|
1366
|
String ref;
|
1367
|
// <Categorical ref="c4">
|
1368
|
List<Element> elCategoricals = elSummaryData.getChildren("Categorical", sddNamespace);
|
1369
|
int k = 0;
|
1370
|
//for each Categorical
|
1371
|
for (Element elCategorical : elCategoricals){
|
1372
|
if ((++k % modCount) == 0){ logger.warn("Categorical handled: " + (k-1));}
|
1373
|
ref = elCategorical.getAttributeValue("ref");
|
1374
|
Feature feature = features.get(ref);
|
1375
|
CategoricalData categoricalData = CategoricalData.NewInstance();
|
1376
|
categoricalData.setFeature(feature);
|
1377
|
|
1378
|
// <State ref="s3"/>
|
1379
|
List<Element> elStates = elCategorical.getChildren("State", sddNamespace);
|
1380
|
int l = 0;
|
1381
|
|
1382
|
//for each State
|
1383
|
for (Element elState : elStates){
|
1384
|
if ((++l % modCount) == 0){ logger.info("States handled: " + (l-1));}
|
1385
|
ref = elState.getAttributeValue("ref");
|
1386
|
State state = states.get(ref);
|
1387
|
if (state != null) {
|
1388
|
StateData stateData = StateData.NewInstance();
|
1389
|
stateData.setState(state);
|
1390
|
List<Element> elModifiers = elState.getChildren("Modifier", sddNamespace);
|
1391
|
for (Element elModifier : elModifiers){
|
1392
|
ref = elModifier.getAttributeValue("ref");
|
1393
|
Modifier modifier = modifiers.get(ref);
|
1394
|
if (modifier != null) {
|
1395
|
stateData.addModifier(modifier);
|
1396
|
}
|
1397
|
}
|
1398
|
categoricalData.addState(stateData);
|
1399
|
}
|
1400
|
taxonDescription.addElement(categoricalData);
|
1401
|
}
|
1402
|
}
|
1403
|
}
|
1404
|
|
1405
|
// imports the persons associated with the dataset creation, modification, related publications
|
1406
|
protected void importAgents(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1407
|
// <Agents>
|
1408
|
logger.info("start Agents ...");
|
1409
|
Element elAgents = elDataset.getChild("Agents",sddNamespace);
|
1410
|
if (elAgents != null) {
|
1411
|
// <Agent id="a1">
|
1412
|
List <Element> listAgents = elAgents.getChildren("Agent", sddNamespace);
|
1413
|
int j = 0;
|
1414
|
//for each Agent
|
1415
|
for (Element elAgent : listAgents){
|
1416
|
|
1417
|
try {
|
1418
|
|
1419
|
String idA = elAgent.getAttributeValue("id");
|
1420
|
|
1421
|
// <Representation>
|
1422
|
// <Label>Kevin Thiele</Label>
|
1423
|
// <Detail role="Description">Ali Baba is also known as r.a.m.</Detail>
|
1424
|
// </Representation>
|
1425
|
Person person = Person.NewInstance();
|
1426
|
importRepresentation(elAgent, sddNamespace, person, idA, cdmState);
|
1427
|
person.addSource(IdentifiableSource.NewInstance(idA, "Agent"));
|
1428
|
|
1429
|
/*XIM <Links>
|
1430
|
Element elLinks = elAgent.getChild("Links",sddNamespace);
|
1431
|
|
1432
|
if (elLinks != null) {
|
1433
|
|
1434
|
// <Link rel="Alternate" href="http://www.diversitycampus.net/people/hagedorn"/>
|
1435
|
List<Element> listLinks = elLinks.getChildren("Link", sddNamespace);
|
1436
|
int k = 0;
|
1437
|
//for each Link
|
1438
|
for (Element elLink : listLinks){
|
1439
|
|
1440
|
try {
|
1441
|
|
1442
|
String rel = elLink.getAttributeValue("rel");
|
1443
|
String href = elLink.getAttributeValue("href");
|
1444
|
|
1445
|
Media link = Media.NewInstance();
|
1446
|
MediaRepresentation mr = MediaRepresentation.NewInstance();
|
1447
|
mr.addRepresentationPart(MediaRepresentationPart.NewInstance(href, null));
|
1448
|
link.addRepresentation(mr);
|
1449
|
person.addMedia(link);
|
1450
|
|
1451
|
} catch (Exception e) {
|
1452
|
//FIXME
|
1453
|
logger.warn("Import of Link " + k + " failed.");
|
1454
|
success = false;
|
1455
|
}
|
1456
|
|
1457
|
if ((++k % modCount) == 0){ logger.info("Links handled: " + k);}
|
1458
|
|
1459
|
}
|
1460
|
}
|
1461
|
*/
|
1462
|
if (authors.containsKey(idA)) {
|
1463
|
authors.put(idA,person);
|
1464
|
}
|
1465
|
|
1466
|
if (editors.containsKey(idA)) {
|
1467
|
editors.put(idA, person);
|
1468
|
}
|
1469
|
|
1470
|
} catch (Exception e) {
|
1471
|
//FIXME
|
1472
|
logger.warn("Import of Agent " + j + " failed.");
|
1473
|
cdmState.setUnsuccessfull();
|
1474
|
}
|
1475
|
|
1476
|
if ((++j % modCount) == 0){ logger.info("Agents handled: " + j);}
|
1477
|
|
1478
|
}
|
1479
|
}
|
1480
|
}
|
1481
|
|
1482
|
// imports publications related with the data set
|
1483
|
protected void importPublications(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1484
|
/* <Publications>
|
1485
|
<Publication id="p112">
|
1486
|
<Representation>
|
1487
|
<Label>Gee, X. & Haa, Y. (2003). How to be happy in five minutes. Instant Gratifications, Palm Beach.</Label>
|
1488
|
</Representation>
|
1489
|
<Links>
|
1490
|
<Link rel="BasedOn" href="doi:10.1992/32311"/>
|
1491
|
<Link rel="Alternate" href="http://some.service.net/providing/bibliographic.data"/>
|
1492
|
</Links>
|
1493
|
</Publications>
|
1494
|
*/
|
1495
|
logger.info("start Publications ...");
|
1496
|
Element elPublications = elDataset.getChild("Publications",sddNamespace);
|
1497
|
|
1498
|
if (elPublications != null) {
|
1499
|
List<Element> listPublications = elPublications.getChildren("Publication", sddNamespace);
|
1500
|
int j = 0;
|
1501
|
for (Element elPublication : listPublications){
|
1502
|
|
1503
|
try {
|
1504
|
|
1505
|
String idP = elPublication.getAttributeValue("id");
|
1506
|
Reference publication = ReferenceFactory.newArticle();
|
1507
|
importRepresentation(elPublication, sddNamespace, publication, idP, cdmState);
|
1508
|
|
1509
|
publications.put(idP,publication);
|
1510
|
|
1511
|
} catch (Exception e) {
|
1512
|
logger.warn("Import of Publication " + j + " failed.");
|
1513
|
cdmState.setUnsuccessfull();
|
1514
|
}
|
1515
|
|
1516
|
if ((++j % modCount) == 0){ logger.info("Publications handled: " + j);}
|
1517
|
|
1518
|
}
|
1519
|
}
|
1520
|
}
|
1521
|
|
1522
|
// imports media objects such as images //FIXME check mediaobj
|
1523
|
protected void importMediaObjects(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1524
|
// <MediaObjects>
|
1525
|
logger.info("start MediaObjects ...");
|
1526
|
Element elMediaObjects = elDataset.getChild("MediaObjects",sddNamespace);
|
1527
|
|
1528
|
if (elMediaObjects != null) {
|
1529
|
// <MediaObject id="m1">
|
1530
|
List<Element> listMediaObjects = elMediaObjects.getChildren("MediaObject", sddNamespace);
|
1531
|
int j = 0;
|
1532
|
for (Element elMO : listMediaObjects){
|
1533
|
|
1534
|
String id = "";
|
1535
|
|
1536
|
try {
|
1537
|
String idMO = elMO.getAttributeValue("id");
|
1538
|
id = idMO;
|
1539
|
|
1540
|
// <Representation>
|
1541
|
// <Label>Image description, e.g. to be used for alt-attribute in html.</Label>
|
1542
|
// </Representation>
|
1543
|
Media media = Media.NewInstance();
|
1544
|
importRepresentation(elMO, sddNamespace, media, idMO, cdmState);
|
1545
|
|
1546
|
// <Type>Image</Type>
|
1547
|
// <Source href="http://test.edu/test.jpg"/>
|
1548
|
String type = (String)ImportHelper.getXmlInputValue(elMO,"Type",sddNamespace);
|
1549
|
|
1550
|
if ((type != null) && (type.equals("Image"))) {
|
1551
|
Element elSource = elMO.getChild("Source",sddNamespace);
|
1552
|
String href = elSource.getAttributeValue("href");
|
1553
|
|
1554
|
ImageInfo imageMetaData = null;
|
1555
|
ImageFile image = null;
|
1556
|
if (href.substring(0,7).equals("http://")) {
|
1557
|
try{
|
1558
|
URL url = new URL(href);
|
1559
|
|
1560
|
imageMetaData = ImageInfo.NewInstance(url.toURI(), 0);
|
1561
|
image = ImageFile.NewInstance(url.toURI(), null, imageMetaData);
|
1562
|
} catch (MalformedURLException e) {
|
1563
|
logger.error("Malformed URL", e);
|
1564
|
}
|
1565
|
} else {
|
1566
|
String sns = cdmState.getConfig().getSourceNameString();
|
1567
|
File f = new File(sns);
|
1568
|
File parent = f.getParentFile();
|
1569
|
String fi = parent.toString() + File.separator + href;
|
1570
|
File file = new File(fi);
|
1571
|
imageMetaData = ImageInfo.NewInstance(new URI(fi), 0); //file
|
1572
|
image = ImageFile.NewInstance(file.toURI(), null, imageMetaData);
|
1573
|
}
|
1574
|
MediaRepresentation representation = MediaRepresentation.NewInstance(imageMetaData.getMimeType(), null);
|
1575
|
representation.addRepresentationPart(image);
|
1576
|
|
1577
|
media.addRepresentation(representation);
|
1578
|
|
1579
|
ArrayList<CdmBase> lcb = (ArrayList<CdmBase>) mediaObject_ListCdmBase.get(idMO);
|
1580
|
if (lcb != null) {
|
1581
|
for (int k = 0; k < lcb.size(); k++) {
|
1582
|
if (lcb.get(k) instanceof DefinedTermBase) {
|
1583
|
DefinedTermBase dtb = (DefinedTermBase) lcb.get(k);
|
1584
|
// if (lcb.get(0) instanceof DefinedTermBase) {
|
1585
|
// DefinedTermBase dtb = (DefinedTermBase) lcb.get(0);
|
1586
|
// if (dtb!=null) {
|
1587
|
// if (k == 0) {
|
1588
|
dtb.addMedia(media);
|
1589
|
//System.out.println(dtb.getLabel());
|
1590
|
// } else {
|
1591
|
// Media me = (Media) media.clone();
|
1592
|
// dtb.addMedia(me);
|
1593
|
// }
|
1594
|
// }
|
1595
|
} else if (lcb.get(k) instanceof Reference) {
|
1596
|
Reference rb = (Reference) lcb.get(k);
|
1597
|
//} else if (lcb.get(0) instanceof Reference) {
|
1598
|
//Reference rb = (Reference) lcb.get(0);
|
1599
|
// rb.setTitleCache(label);
|
1600
|
// if (rb!=null) {
|
1601
|
// if (k == 0) {
|
1602
|
rb.addMedia(media);
|
1603
|
//System.out.println(rb.getTitle());
|
1604
|
// } else {
|
1605
|
// Media me = (Media) media.clone();
|
1606
|
// rb.addMedia(me);
|
1607
|
// }
|
1608
|
// }
|
1609
|
// else if (lcb.get(k) instanceof TaxonNameBase){
|
1610
|
// TaxonNameBase tb = (TaxonNameBase) lcb.get(k);
|
1611
|
// tb.addMedia(media);
|
1612
|
} else {
|
1613
|
logger.warn("Can't handle associated media for " + lcb.get(k).getId() + "(" + lcb.get(k).getClass().getSimpleName()+")" );
|
1614
|
}
|
1615
|
}
|
1616
|
}
|
1617
|
}
|
1618
|
|
1619
|
} catch (Exception e) {
|
1620
|
//FIXME
|
1621
|
logger.warn("Could not attach MediaObject " + j + "(SDD: " + id + ") to several objects.");
|
1622
|
cdmState.setUnsuccessfull();
|
1623
|
}
|
1624
|
|
1625
|
if ((++j % modCount) == 0){ logger.info("MediaObjects handled: " + j);
|
1626
|
|
1627
|
}
|
1628
|
}
|
1629
|
}
|
1630
|
}
|
1631
|
|
1632
|
// imports the <DescriptiveConcepts> block ; DescriptiveConcepts are used as nodes in CharacterTrees and Characters as leaves
|
1633
|
// but since Modifiers can be linked to DescriptiveConcepts they are stored as features with a particular Marker
|
1634
|
protected void importDescriptiveConcepts(Element elDataset, Namespace sddNamespace, SDDImportState state){
|
1635
|
/* <DescriptiveConcepts>
|
1636
|
<DescriptiveConcept id="dc0">
|
1637
|
<Representation>
|
1638
|
<Label>Fixed set of modifiers supported in Lucid3</Label>
|
1639
|
</Representation>
|
1640
|
<Modifiers>
|
1641
|
<Modifier id="mod1">
|
1642
|
<Representation>
|
1643
|
<Label>rarely</Label>
|
1644
|
</Representation>
|
1645
|
<ModifierClass>Frequency</ModifierClass>
|
1646
|
<ProportionRange lowerestimate="0.0" upperestimate="0.25"/>
|
1647
|
</Modifier>
|
1648
|
</Modifiers>
|
1649
|
</DescriptiveConcept>
|
1650
|
</DescriptiveConcepts>
|
1651
|
*/
|
1652
|
logger.info("start DescriptiveConcepts ...");
|
1653
|
Element elDescriptiveConcepts = elDataset.getChild("DescriptiveConcepts",sddNamespace);
|
1654
|
if (elDescriptiveConcepts != null) {
|
1655
|
List<Element> listDescriptiveConcepts = elDescriptiveConcepts.getChildren("DescriptiveConcept", sddNamespace);
|
1656
|
int j = 0;
|
1657
|
|
1658
|
for (Element elDescriptiveConcept : listDescriptiveConcepts){
|
1659
|
try {
|
1660
|
String id = elDescriptiveConcept.getAttributeValue("id");
|
1661
|
Feature feature = Feature.NewInstance();
|
1662
|
feature.setKindOf(Feature.DESCRIPTION());
|
1663
|
if (!id.equals("")) {
|
1664
|
// <Representation>
|
1665
|
// <Label>Body</Label>
|
1666
|
importRepresentation(elDescriptiveConcept, sddNamespace, feature, id, state);
|
1667
|
features.put(id, feature);
|
1668
|
getTermService().save(feature);//XIM
|
1669
|
descriptiveConcepts.add(feature);
|
1670
|
// imports the modifiers
|
1671
|
Element elModifiers = elDescriptiveConcept.getChild("Modifiers", sddNamespace);
|
1672
|
if (elModifiers !=null){
|
1673
|
List<Element> listModifiers = elModifiers.getChildren("Modifier", sddNamespace);
|
1674
|
TermVocabulary<Modifier> termVocabularyState = TermVocabulary.NewInstance(null, null, null, null);
|
1675
|
for (Element elModifier : listModifiers) {
|
1676
|
Modifier modif = Modifier.NewInstance();
|
1677
|
String idmod = elModifier.getAttributeValue("id");
|
1678
|
importRepresentation(elModifier, sddNamespace, modif, idmod, state);
|
1679
|
termVocabularyState.addTerm(modif);
|
1680
|
//termVocabularyStates.add(termVocabularyState);
|
1681
|
getVocabularyService().save(termVocabularyState);//XIM
|
1682
|
modifiers.put(idmod, modif);
|
1683
|
}
|
1684
|
feature.addRecommendedModifierEnumeration(termVocabularyState);
|
1685
|
}
|
1686
|
|
1687
|
}
|
1688
|
}
|
1689
|
catch (Exception e) {
|
1690
|
logger.warn("Import of DescriptiveConcept " + j + " failed: " + e.getMessage());
|
1691
|
}
|
1692
|
if ((++j % modCount) == 0){ logger.info("DescriptiveConcepts handled: " + j);}
|
1693
|
|
1694
|
}
|
1695
|
}
|
1696
|
}
|
1697
|
|
1698
|
// imports the <CharacterTrees> block
|
1699
|
protected void importCharacterTrees(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1700
|
// <CharacterTrees>
|
1701
|
logger.info("start CharacterTrees ...");
|
1702
|
Element elCharacterTrees = elDataset.getChild("CharacterTrees",sddNamespace);
|
1703
|
|
1704
|
if (elCharacterTrees != null) {
|
1705
|
List<Element> listCharacterTrees = elCharacterTrees.getChildren("CharacterTree", sddNamespace);
|
1706
|
int j = 0;
|
1707
|
for (Element elCharacterTree : listCharacterTrees){
|
1708
|
try {
|
1709
|
Element elRepresentation = elCharacterTree.getChild("Representation",sddNamespace);
|
1710
|
String label = (String)ImportHelper.getXmlInputValue(elRepresentation,"Label",sddNamespace);
|
1711
|
//Element elDesignedFor = elCharacterTree.getChild("DesignedFor",sddNamespace);//TODO ?
|
1712
|
|
1713
|
FeatureTree featureTree = FeatureTree.NewInstance();
|
1714
|
importRepresentation(elCharacterTree, sddNamespace, featureTree, "", cdmState);
|
1715
|
FeatureNode root = featureTree.getRoot();
|
1716
|
List<Element> listeOfNodes = elCharacterTree.getChildren("Nodes", sddNamespace);
|
1717
|
|
1718
|
//Nodes of CharacterTrees in SDD always refer to DescriptiveConcepts
|
1719
|
for (Element elNodes : listeOfNodes) {
|
1720
|
handleCharacterNodes(sddNamespace, root, elNodes);
|
1721
|
}
|
1722
|
featureTrees.add(featureTree);
|
1723
|
if (workingSet.getDescriptiveSystem() != null){
|
1724
|
//TODO how to handle multiple
|
1725
|
logger.warn("Multiple feature trees not yet supported");
|
1726
|
}else{
|
1727
|
workingSet.setDescriptiveSystem(featureTree);
|
1728
|
}
|
1729
|
}
|
1730
|
|
1731
|
catch (Exception e) {
|
1732
|
logger.warn("Import of Character tree " + j + " failed.");
|
1733
|
cdmState.setUnsuccessfull();
|
1734
|
}
|
1735
|
if ((++j % modCount) == 0){ logger.info("CharacterTrees handled: " + j);}
|
1736
|
|
1737
|
}
|
1738
|
|
1739
|
}
|
1740
|
}
|
1741
|
|
1742
|
/**
|
1743
|
* @param sddNamespace
|
1744
|
* @param root
|
1745
|
* @param elNodes
|
1746
|
*/
|
1747
|
private void handleCharacterNodes(Namespace sddNamespace, FeatureNode root, Element elNodes) {
|
1748
|
List<Element> listNodes = elNodes.getChildren("Node", sddNamespace);
|
1749
|
if (listNodes != null) {
|
1750
|
for (Element elNode : listNodes){
|
1751
|
String idN = elNode.getAttributeValue("id");
|
1752
|
FeatureNode fn = null;
|
1753
|
Feature dc = null;
|
1754
|
if (idN!=null) {
|
1755
|
// DescriptiveConcepts are used as nodes in CharacterTrees
|
1756
|
Element elDescriptiveConcept = elNode.getChild("DescriptiveConcept", sddNamespace);
|
1757
|
if (elDescriptiveConcept != null){
|
1758
|
String refDC = elDescriptiveConcept.getAttributeValue("ref");
|
1759
|
dc = features.get(refDC);
|
1760
|
fn = FeatureNode.NewInstance(dc);
|
1761
|
}
|
1762
|
if (fn==null){
|
1763
|
fn = FeatureNode.NewInstance();
|
1764
|
}
|
1765
|
Element elParent = elNode.getChild("Parent", sddNamespace);
|
1766
|
// in SDD links between Nodes are referenced by the <Parent> tag
|
1767
|
if (elParent!=null){
|
1768
|
String refP = elParent.getAttributeValue("ref");
|
1769
|
if (refP!=null) {
|
1770
|
FeatureNode parent = featureNodes.get(refP);
|
1771
|
if (parent==null){
|
1772
|
root.addChild(fn); // if no parent found or the reference is broken, add the node to the root of the tree
|
1773
|
}
|
1774
|
else {
|
1775
|
parent.addChild(fn);
|
1776
|
}
|
1777
|
}
|
1778
|
}
|
1779
|
else {
|
1780
|
root.addChild(fn); // if no parent found or the reference is broken, add the node to the root of the tree
|
1781
|
}
|
1782
|
}
|
1783
|
featureNodes.put(idN, fn);
|
1784
|
}
|
1785
|
}
|
1786
|
|
1787
|
// Leaves of CharacterTrees in SDD are always CharNodes (referring to Characters)
|
1788
|
List<Element> listCharNodes = elNodes.getChildren("CharNode", sddNamespace);
|
1789
|
if (listCharNodes != null) {
|
1790
|
for (Element elCharNode : listCharNodes){
|
1791
|
Element elParent = elCharNode.getChild("Parent", sddNamespace);
|
1792
|
Element elCharacter = elCharNode.getChild("Character", sddNamespace);
|
1793
|
Element elDependencyRules = elCharNode.getChild("DependencyRules", sddNamespace);
|
1794
|
FeatureNode fn = FeatureNode.NewInstance();
|
1795
|
|
1796
|
if (elDependencyRules!=null){
|
1797
|
Element elInapplicableIf = elCharNode.getChild("InapplicableIf", sddNamespace);
|
1798
|
if (elInapplicableIf!=null){
|
1799
|
List<Element> listStates = elInapplicableIf.getChildren("State", sddNamespace);
|
1800
|
for (Element stateElement : listStates) {
|
1801
|
String refState = stateElement.getAttributeValue("ref");
|
1802
|
if ((refState!=null)&&(!refState.equals(""))) {
|
1803
|
State state = states.get(refState);
|
1804
|
fn.addInapplicableState(state);
|
1805
|
}
|
1806
|
}
|
1807
|
}
|
1808
|
Element elOnlyapplicableIf = elCharNode.getChild("OnlyApplicableIf", sddNamespace);
|
1809
|
if (elOnlyapplicableIf!=null){
|
1810
|
List<Element> listStates = elInapplicableIf.getChildren("State", sddNamespace);
|
1811
|
for (Element stateElement : listStates) {
|
1812
|
String refState = stateElement.getAttributeValue("ref");
|
1813
|
if ((refState!=null)&&(!refState.equals(""))) {
|
1814
|
State state = states.get(refState);
|
1815
|
fn.addApplicableState(state);
|
1816
|
}
|
1817
|
}
|
1818
|
}
|
1819
|
}
|
1820
|
|
1821
|
if (elParent!=null){
|
1822
|
String refP = elParent.getAttributeValue("ref");
|
1823
|
if ((refP!=null)&&(!refP.equals(""))) {
|
1824
|
FeatureNode parent = featureNodes.get(refP);
|
1825
|
if (parent==null){
|
1826
|
parent = root; // if no parent found or the reference is broken, add the node to the root of the tree
|
1827
|
}
|
1828
|
parent.addChild(fn);
|
1829
|
}
|
1830
|
}
|
1831
|
String refC = elCharacter.getAttributeValue("ref");
|
1832
|
if ((refC!=null)&&(!refC.equals(""))){
|
1833
|
Feature character = features.get(refC);
|
1834
|
fn.setFeature(character);
|
1835
|
featureNodes.put(refC, fn);
|
1836
|
}
|
1837
|
}
|
1838
|
}
|
1839
|
}
|
1840
|
|
1841
|
// imports the <TaxonHierarchies> block
|
1842
|
protected void importTaxonHierarchies(Element elDataset, Namespace sddNamespace, SDDImportState cdmState){
|
1843
|
|
1844
|
logger.info("start TaxonHierarchies ...");
|
1845
|
Element elTaxonHierarchies = elDataset.getChild("TaxonHierarchies",sddNamespace);
|
1846
|
|
1847
|
if (elTaxonHierarchies != null) {
|
1848
|
List<Element> listTaxonHierarchies = elTaxonHierarchies.getChildren("TaxonHierarchy", sddNamespace);
|
1849
|
int j = 0;
|
1850
|
for (Element elTaxonHierarchy : listTaxonHierarchies){
|
1851
|
try {
|
1852
|
Element elRepresentation = elTaxonHierarchy.getChild("Representation",sddNamespace);
|
1853
|
String label = (String)ImportHelper.getXmlInputValue(elRepresentation,"Label",sddNamespace);
|
1854
|
Classification classification = Classification.NewInstance(label);
|
1855
|
importRepresentation(elTaxonHierarchy, sddNamespace, classification, "", cdmState);
|
1856
|
|
1857
|
Set<TaxonNode> root = classification.getChildNodes();
|
1858
|
Element elNodes = elTaxonHierarchy.getChild("Nodes", sddNamespace); // There can be only one <Nodes> block for TaxonHierarchies
|
1859
|
List<Element> listNodes = elNodes.getChildren("Node", sddNamespace);
|
1860
|
|
1861
|
for (Element elNode : listNodes){
|
1862
|
String idN = elNode.getAttributeValue("id");
|
1863
|
TaxonNameBase tnb = null;
|
1864
|
if (!idN.equals("")) {
|
1865
|
Element elTaxonName = elNode.getChild("TaxonName", sddNamespace);
|
1866
|
String refTN = elTaxonName.getAttributeValue("ref");
|
1867
|
tnb = taxonNameBases.get(refTN);
|
1868
|
Taxon taxon = (Taxon) tnb.getTaxa().iterator().next() ;
|
1869
|
Element elParent = elNode.getChild("Parent", sddNamespace);
|
1870
|
if (elParent!=null){
|
1871
|
String refP = elParent.getAttributeValue("ref");
|
1872
|
if (!refP.equals("")) {
|
1873
|
TaxonNode parent = taxonNodes.get(refP);
|
1874
|
TaxonNode child = parent.addChildTaxon(taxon, sec, "", Synonym.NewInstance(tnb, sec));
|
1875
|
taxonNodes.put(idN,child);
|
1876
|
}
|
1877
|
}
|
1878
|
else {
|
1879
|
TaxonNode tn = classification.addChildTaxon(taxon, sec, "", Synonym.NewInstance(tnb, sec)); // if no parent found or the reference is broken, add the node to the root of the tree
|
1880
|
taxonNodes.put(idN,tn);
|
1881
|
}
|
1882
|
}
|
1883
|
}
|
1884
|
|
1885
|
classifications.add(classification);
|
1886
|
}
|
1887
|
|
1888
|
catch (Exception e) {
|
1889
|
//FIXME
|
1890
|
logger.warn("Import of Taxon Hierarchy " + j + " failed.");
|
1891
|
cdmState.setUnsuccessfull();
|
1892
|
}
|
1893
|
|
1894
|
if ((++j % modCount) == 0){ logger.info("TaxonHierarchies handled: " + j);}
|
1895
|
|
1896
|
}
|
1897
|
|
1898
|
}
|
1899
|
}
|
1900
|
|
1901
|
|
1902
|
// imports the <GeographicAreas> block
|
1903
|
protected void importGeographicAreas(Element elDataset, Namespace sddNamespace, SDDImportState cdmState) {
|
1904
|
Element elGeographicAreas = elDataset.getChild("GeographicAreas",sddNamespace);
|
1905
|
if (elGeographicAreas != null) {
|
1906
|
List<Element> listGeographicAreas = elGeographicAreas.getChildren("GeographicArea", sddNamespace);
|
1907
|
int j = 0;
|
1908
|
|
1909
|
for (Element elGeographicArea : listGeographicAreas){
|
1910
|
|
1911
|
String id = elGeographicArea.getAttributeValue("id");
|
1912
|
NamedArea na = new NamedArea();
|
1913
|
importRepresentation(elGeographicArea, sddNamespace, na, id, cdmState);
|
1914
|
namedAreas.put(id,na);
|
1915
|
}
|
1916
|
if ((++j % modCount) == 0){ logger.info("GeographicAreas handled: " + j);}
|
1917
|
}
|
1918
|
}
|
1919
|
}
|