|
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.markup;
|
|
11
|
|
|
12
|
import java.util.ArrayList;
|
|
13
|
import java.util.List;
|
|
14
|
import java.util.Map;
|
|
15
|
import java.util.Set;
|
|
16
|
import java.util.UUID;
|
|
17
|
import java.util.regex.Matcher;
|
|
18
|
import java.util.regex.Pattern;
|
|
19
|
|
|
20
|
import javax.xml.stream.XMLEventReader;
|
|
21
|
import javax.xml.stream.XMLStreamException;
|
|
22
|
import javax.xml.stream.events.Attribute;
|
|
23
|
import javax.xml.stream.events.XMLEvent;
|
|
24
|
|
|
25
|
import org.apache.commons.lang.StringUtils;
|
|
26
|
import org.apache.logging.log4j.LogManager;import org.apache.logging.log4j.Logger;
|
|
27
|
|
|
28
|
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacade;
|
|
29
|
import eu.etaxonomy.cdm.api.facade.DerivedUnitFacadeCacheStrategy;
|
|
30
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
31
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
|
32
|
import eu.etaxonomy.cdm.model.common.AnnotatableEntity;
|
|
33
|
import eu.etaxonomy.cdm.model.common.Annotation;
|
|
34
|
import eu.etaxonomy.cdm.model.common.AnnotationType;
|
|
35
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
36
|
import eu.etaxonomy.cdm.model.common.Language;
|
|
37
|
import eu.etaxonomy.cdm.model.common.Marker;
|
|
38
|
import eu.etaxonomy.cdm.model.common.MarkerType;
|
|
39
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
|
40
|
import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
|
|
41
|
import eu.etaxonomy.cdm.model.description.Feature;
|
|
42
|
import eu.etaxonomy.cdm.model.description.IndividualsAssociation;
|
|
43
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
|
44
|
import eu.etaxonomy.cdm.model.location.Country;
|
|
45
|
import eu.etaxonomy.cdm.model.location.NamedArea;
|
|
46
|
import eu.etaxonomy.cdm.model.location.NamedAreaLevel;
|
|
47
|
import eu.etaxonomy.cdm.model.name.HomotypicalGroup;
|
|
48
|
import eu.etaxonomy.cdm.model.name.INonViralName;
|
|
49
|
import eu.etaxonomy.cdm.model.name.Rank;
|
|
50
|
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignation;
|
|
51
|
import eu.etaxonomy.cdm.model.name.SpecimenTypeDesignationStatus;
|
|
52
|
import eu.etaxonomy.cdm.model.name.TaxonName;
|
|
53
|
import eu.etaxonomy.cdm.model.occurrence.Collection;
|
|
54
|
import eu.etaxonomy.cdm.model.occurrence.DerivedUnit;
|
|
55
|
import eu.etaxonomy.cdm.model.occurrence.DeterminationEvent;
|
|
56
|
import eu.etaxonomy.cdm.model.occurrence.FieldUnit;
|
|
57
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase;
|
|
58
|
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationType;
|
|
59
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
60
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
|
61
|
import eu.etaxonomy.cdm.model.term.DefinedTerm;
|
|
62
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
|
63
|
import eu.etaxonomy.cdm.strategy.parser.SpecimenTypeParser;
|
|
64
|
import eu.etaxonomy.cdm.strategy.parser.SpecimenTypeParser.TypeInfo;
|
|
65
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
|
66
|
|
|
67
|
/**
|
|
68
|
* @author a.mueller
|
|
69
|
* @since 30.05.2012
|
|
70
|
*
|
|
71
|
*/
|
|
72
|
public class MarkupSpecimenImport extends MarkupImportBase {
|
|
73
|
@SuppressWarnings("unused")
|
|
74
|
private static final Logger logger = LogManager.getLogger(MarkupSpecimenImport.class);
|
|
75
|
|
|
76
|
private static final String ALTERNATIVE_COLLECTION_TYPE_STATUS = "alternativeCollectionTypeStatus";
|
|
77
|
private static final String ALTERNATIVE_COLLECTOR = "alternativeCollector";
|
|
78
|
private static final String ALTERNATIVE_FIELD_NUM = "alternativeFieldNum";
|
|
79
|
private static final String COLLECTOR = "collector";
|
|
80
|
private static final String COLLECTION = "collection";
|
|
81
|
private static final String COLLECTION_AND_TYPE = "collectionAndType";
|
|
82
|
private static final String COLLECTION_TYPE_STATUS = "collectionTypeStatus";
|
|
83
|
private static final String DAY = "day";
|
|
84
|
private static final String DESTROYED = "destroyed";
|
|
85
|
private static final String FIELD_NUM = "fieldNum";
|
|
86
|
private static final String FULL_TYPE = "fullType";
|
|
87
|
private static final String FULL_DATE = "fullDate";
|
|
88
|
private static final String GATHERING_NOTES = "gatheringNotes";
|
|
89
|
private static final String LOST = "lost";
|
|
90
|
private static final String MONTH = "month";
|
|
91
|
private static final String SUB_GATHERING = "subGathering";
|
|
92
|
private static final String NOT_FOUND = "notFound";
|
|
93
|
private static final String NOT_SEEN = "notSeen";
|
|
94
|
private static final String ORIGINAL_DETERMINATION = "originalDetermination";
|
|
95
|
|
|
96
|
private static final String UNKNOWN = "unknown";
|
|
97
|
private static final String YEAR = "year";
|
|
98
|
|
|
99
|
|
|
100
|
|
|
101
|
public MarkupSpecimenImport(MarkupDocumentImport docImport) {
|
|
102
|
super(docImport);
|
|
103
|
}
|
|
104
|
|
|
105
|
|
|
106
|
public void handleSpecimenType(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent,
|
|
107
|
HomotypicalGroup homotypicalGroup) throws XMLStreamException {
|
|
108
|
|
|
109
|
// attributes
|
|
110
|
Map<String, Attribute> attributes = getAttributes(parentEvent);
|
|
111
|
String typeStatus = getAndRemoveAttributeValue(attributes, TYPE_STATUS);
|
|
112
|
String notSeen = getAndRemoveAttributeValue(attributes, NOT_SEEN);
|
|
113
|
String unknown = getAndRemoveAttributeValue(attributes, UNKNOWN);
|
|
114
|
String notFound = getAndRemoveAttributeValue(attributes, NOT_FOUND);
|
|
115
|
String destroyed = getAndRemoveAttributeValue(attributes, DESTROYED);
|
|
116
|
String lost = getAndRemoveAttributeValue(attributes, LOST);
|
|
117
|
checkNoAttributes(attributes, parentEvent);
|
|
118
|
if (isNotBlank(typeStatus)) {
|
|
119
|
// TODO
|
|
120
|
// currently not needed
|
|
121
|
fireWarningEvent("Type status not yet used", parentEvent, 4);
|
|
122
|
}
|
|
123
|
if (isNotBlank(notSeen)) {
|
|
124
|
handleNotYetImplementedAttribute(attributes, NOT_SEEN, parentEvent);
|
|
125
|
}
|
|
126
|
if (isNotBlank(unknown)) {
|
|
127
|
handleNotYetImplementedAttribute(attributes, UNKNOWN, parentEvent);
|
|
128
|
}
|
|
129
|
if (isNotBlank(notFound)) {
|
|
130
|
handleNotYetImplementedAttribute(attributes, NOT_FOUND, parentEvent);
|
|
131
|
}
|
|
132
|
if (isNotBlank(destroyed)) {
|
|
133
|
handleNotYetImplementedAttribute(attributes, DESTROYED, parentEvent);
|
|
134
|
}
|
|
135
|
if (isNotBlank(lost)) {
|
|
136
|
handleNotYetImplementedAttribute(attributes, LOST, parentEvent);
|
|
137
|
}
|
|
138
|
|
|
139
|
INonViralName givenName = null;
|
|
140
|
Set<TaxonName> names = homotypicalGroup.getTypifiedNames();
|
|
141
|
if (names.isEmpty()) {
|
|
142
|
String message = "There is no name in a homotypical group. Can't create the specimen type";
|
|
143
|
fireWarningEvent(message, parentEvent, 8);
|
|
144
|
} else {
|
|
145
|
givenName = CdmBase.deproxy(names.iterator().next());
|
|
146
|
}
|
|
147
|
|
|
148
|
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(SpecimenOrObservationType.PreservedSpecimen);
|
|
149
|
state.setFirstSpecimenInFacade(true);
|
|
150
|
String text = "";
|
|
151
|
state.resetCollectionAndType();
|
|
152
|
state.setSpecimenType(true);
|
|
153
|
boolean isFullType = false;
|
|
154
|
// elements
|
|
155
|
while (reader.hasNext()) {
|
|
156
|
XMLEvent next = readNoWhitespace(reader);
|
|
157
|
if (isMyEndingElement(next, parentEvent)) {
|
|
158
|
if (! isFullType){
|
|
159
|
makeSpecimenType(state, facade, text, state.getCollectionAndType(), givenName, parentEvent);
|
|
160
|
}
|
|
161
|
state.setSpecimenType(false);
|
|
162
|
state.resetCollectionAndType();
|
|
163
|
state.setFirstSpecimenInFacade(false);
|
|
164
|
return;
|
|
165
|
} else if (isStartingElement(next, FULL_TYPE)) {
|
|
166
|
handleAmbigousManually(state, reader, next.asStartElement());
|
|
167
|
isFullType = true;
|
|
168
|
} else if (isStartingElement(next, TYPE_STATUS)) {
|
|
169
|
handleNotYetImplementedElement(next);
|
|
170
|
} else if (isStartingElement(next, GATHERING)) {
|
|
171
|
handleGathering(state, reader, next, facade);
|
|
172
|
} else if (isStartingElement(next, ORIGINAL_DETERMINATION)) {
|
|
173
|
handleNotYetImplementedElement(next);
|
|
174
|
} else if (isStartingElement(next, SPECIMEN_TYPE)) {
|
|
175
|
handleNotYetImplementedElement(next);
|
|
176
|
} else if (isStartingElement(next, COLLECTION_AND_TYPE)) {
|
|
177
|
String colAndType = getCData(state, reader, next, true);
|
|
178
|
state.addCollectionAndType(colAndType);
|
|
179
|
} else if (isStartingElement(next, CITATION)) {
|
|
180
|
handleNotYetImplementedElement(next);
|
|
181
|
} else if (isStartingElement(next, NOTES)) {
|
|
182
|
handleNotYetImplementedElement(next);
|
|
183
|
} else if (isStartingElement(next, ANNOTATION)) {
|
|
184
|
handleNotYetImplementedElement(next);
|
|
185
|
} else if (next.isCharacters()) {
|
|
186
|
text += next.asCharacters().getData();
|
|
187
|
} else {
|
|
188
|
handleUnexpectedElement(next);
|
|
189
|
}
|
|
190
|
}
|
|
191
|
throw new IllegalStateException("Specimen type has no closing tag");
|
|
192
|
}
|
|
193
|
|
|
194
|
|
|
195
|
|
|
196
|
private void makeSpecimenType(MarkupImportState state, DerivedUnitFacade facade, String text, String collectionAndType,
|
|
197
|
INonViralName name, XMLEvent parentEvent) {
|
|
198
|
text = text.trim();
|
|
199
|
if (isBlank(text) || isPunctuation(text)){
|
|
200
|
//do nothing
|
|
201
|
}else{
|
|
202
|
String message = "Text '%s' not handled for <SpecimenType>";
|
|
203
|
this.fireWarningEvent(String.format(message, text), parentEvent, 4);
|
|
204
|
}
|
|
205
|
|
|
206
|
if (makeFotgSpecimenType(state, collectionAndType, facade, name, parentEvent) || state.getConfig().isUseFotGSpecimenTypeCollectionAndTypeOnly()){
|
|
207
|
return;
|
|
208
|
}else{
|
|
209
|
// remove brackets
|
|
210
|
if (collectionAndType.matches("^\\(.*\\)\\.?$")) {
|
|
211
|
collectionAndType = collectionAndType.replaceAll("\\.$", "");
|
|
212
|
collectionAndType = collectionAndType.substring(1, collectionAndType.length() - 1);
|
|
213
|
}
|
|
214
|
|
|
215
|
String[] splitsSemi = collectionAndType.split("[;]");
|
|
216
|
for (String splitSemi : splitsSemi) {
|
|
217
|
String[] splitKomma = splitSemi.split("[,]");
|
|
218
|
TypeInfo lastTypeInfo = null;
|
|
219
|
for (String str : splitKomma) {
|
|
220
|
str = str.trim();
|
|
221
|
boolean addToAllNamesInGroup = true;
|
|
222
|
TypeInfo typeInfo = makeSpecimenTypeTypeInfo(state, str, lastTypeInfo, parentEvent);
|
|
223
|
SpecimenTypeDesignationStatus typeStatus = typeInfo.status;
|
|
224
|
Collection collection = this.getCollection(state, typeInfo.collectionString);
|
|
225
|
|
|
226
|
// TODO improve cache strategy handling
|
|
227
|
DerivedUnit typeSpecimen;
|
|
228
|
if (state.isFirstSpecimenInFacade()){
|
|
229
|
state.setFirstSpecimenInFacade(false);
|
|
230
|
typeSpecimen = facade.innerDerivedUnit();
|
|
231
|
typeSpecimen.setCollection(collection);
|
|
232
|
}else{
|
|
233
|
typeSpecimen = facade.addDuplicate(collection, null, null, null, null);
|
|
234
|
}
|
|
235
|
typeSpecimen.setCacheStrategy(DerivedUnitFacadeCacheStrategy.NewInstance());
|
|
236
|
name.addSpecimenTypeDesignation(typeSpecimen, typeStatus,
|
|
237
|
null, null, null, typeInfo.notDesignated, addToAllNamesInGroup);
|
|
238
|
handleNotSeen(state, typeSpecimen, typeInfo);
|
|
239
|
lastTypeInfo = typeInfo;
|
|
240
|
}
|
|
241
|
}
|
|
242
|
}
|
|
243
|
}
|
|
244
|
|
|
245
|
private void handleNotSeen(MarkupImportState state, DerivedUnit typeSpecimen, TypeInfo typeInfo) {
|
|
246
|
if (typeInfo.notSeen){
|
|
247
|
String text = "n.v. for " + state.getConfig().getSourceReference().getAbbrevTitleCache();
|
|
248
|
typeSpecimen.addAnnotation(Annotation.NewInstance(text, AnnotationType.EDITORIAL(), getDefaultLanguage(state)));
|
|
249
|
if(state.getConfig().getSpecimenNotSeenMarkerTypeUuid() != null){
|
|
250
|
UUID uuidNotSeenMarker = state.getConfig().getSpecimenNotSeenMarkerTypeUuid();
|
|
251
|
String markerTypeNotSeenLabel = state.getConfig().getSpecimenNotSeenMarkerTypeLabel();
|
|
252
|
markerTypeNotSeenLabel = markerTypeNotSeenLabel == null ? "Not seen" : markerTypeNotSeenLabel;
|
|
253
|
MarkerType notSeenMarkerType = getMarkerType(state, uuidNotSeenMarker, markerTypeNotSeenLabel, markerTypeNotSeenLabel, null, null);
|
|
254
|
Marker marker = Marker.NewInstance(notSeenMarkerType, true);
|
|
255
|
typeSpecimen.addMarker(marker);
|
|
256
|
}
|
|
257
|
}
|
|
258
|
}
|
|
259
|
|
|
260
|
|
|
261
|
private Pattern fotgTypePattern = null;
|
|
262
|
/**
|
|
263
|
* Implemented for Flora of the Guyanas this may include duplicated code from similar places
|
|
264
|
* @param state
|
|
265
|
* @param collectionAndTypeOrig
|
|
266
|
* @param facade
|
|
267
|
* @param name
|
|
268
|
* @param parentEvent
|
|
269
|
* @return
|
|
270
|
*/
|
|
271
|
private boolean makeFotgSpecimenType(MarkupImportState state, final String collectionAndTypeOrig, DerivedUnitFacade facade, INonViralName name, XMLEvent parentEvent) {
|
|
272
|
String collectionAndType = collectionAndTypeOrig;
|
|
273
|
|
|
274
|
String notDesignatedRE = "not\\s+designated";
|
|
275
|
String designatedByRE = "\\s*\\(((designated\\s+by\\s+|according\\s+to\\s+)[^\\)]+|here\\s+designated)\\)";
|
|
276
|
String typesRE = "(holotype|isotypes?|neotype|isoneotype|syntype|lectotype|isolectotypes?|typ\\.\\scons\\.,?)";
|
|
277
|
String collectionRE = "[A-Z\\-]{1,5}!?";
|
|
278
|
String collectionsRE = String.format("%s(,\\s+%s)*",collectionRE, collectionRE);
|
|
279
|
String addInfoRE = "(not\\s+seen|(presumed\\s+)?destroyed)";
|
|
280
|
String singleTypeTypeRE = String.format("(%s\\s)?%s(,\\s+%s)*", typesRE, collectionsRE, addInfoRE);
|
|
281
|
String allTypesRE = String.format("(\\(not\\s+seen\\)|\\(%s([,;]\\s%s)?\\))", singleTypeTypeRE, singleTypeTypeRE);
|
|
282
|
String designatedRE = String.format("%s(%s)?", allTypesRE, designatedByRE);
|
|
283
|
if (fotgTypePattern == null){
|
|
284
|
|
|
285
|
String pattern = String.format("(%s|%s)", notDesignatedRE, designatedRE );
|
|
286
|
fotgTypePattern = Pattern.compile(pattern);
|
|
287
|
}
|
|
288
|
Matcher matcher = fotgTypePattern.matcher(collectionAndType);
|
|
289
|
|
|
290
|
if (matcher.matches()){
|
|
291
|
fireWarningEvent("Try to synchronize type handling (at least creation) with standard type handling. E.g. use TypeInfo and according algorithms", parentEvent, 2);
|
|
292
|
if (collectionAndType.matches(notDesignatedRE)){
|
|
293
|
SpecimenTypeDesignation desig = SpecimenTypeDesignation.NewInstance();
|
|
294
|
desig.setNotDesignated(true);
|
|
295
|
// name.addSpecimenTypeDesignation(typeSpecimen, status, citation, citationMicroReference, originalNameString, isNotDesignated, addToAllHomotypicNames)
|
|
296
|
name.addTypeDesignation(desig, true);
|
|
297
|
}else if(collectionAndType.matches(designatedRE)){
|
|
298
|
String designatedBy = null;
|
|
299
|
Matcher desigMatcher = Pattern.compile(designatedByRE).matcher(collectionAndType);
|
|
300
|
boolean hasDesignatedBy = desigMatcher.find();
|
|
301
|
if (hasDesignatedBy){
|
|
302
|
designatedBy = desigMatcher.group(0);
|
|
303
|
collectionAndType = collectionAndType.replace(designatedBy, "");
|
|
304
|
}
|
|
305
|
|
|
306
|
//remove brackets
|
|
307
|
collectionAndType = collectionAndType.substring(1, collectionAndType.length() -1);
|
|
308
|
List<String> singleTypes = new ArrayList<String>();
|
|
309
|
Pattern singleTypePattern = Pattern.compile("^" + singleTypeTypeRE);
|
|
310
|
matcher = singleTypePattern.matcher(collectionAndType);
|
|
311
|
while (matcher.find()){
|
|
312
|
String match = matcher.group(0);
|
|
313
|
singleTypes.add(match);
|
|
314
|
collectionAndType = collectionAndType.substring(match.length());
|
|
315
|
if (!collectionAndType.isEmpty()){
|
|
316
|
collectionAndType = collectionAndType.substring(1).trim();
|
|
317
|
}else{
|
|
318
|
break;
|
|
319
|
}
|
|
320
|
matcher = singleTypePattern.matcher(collectionAndType);
|
|
321
|
}
|
|
322
|
|
|
323
|
List<SpecimenTypeDesignation> designations = new ArrayList<SpecimenTypeDesignation>();
|
|
324
|
|
|
325
|
//single types
|
|
326
|
for (String singleTypeOrig : singleTypes){
|
|
327
|
String singleType = singleTypeOrig;
|
|
328
|
//type
|
|
329
|
Pattern typePattern = Pattern.compile("^" + typesRE);
|
|
330
|
matcher = typePattern.matcher(singleType);
|
|
331
|
SpecimenTypeDesignationStatus typeStatus = null;
|
|
332
|
if (matcher.find()){
|
|
333
|
String typeStr = matcher.group(0);
|
|
334
|
singleType = singleType.substring(typeStr.length()).trim();
|
|
335
|
try {
|
|
336
|
typeStatus = SpecimenTypeParser.parseSpecimenTypeStatus(typeStr);
|
|
337
|
} catch (UnknownCdmTypeException e) {
|
|
338
|
fireWarningEvent("specimen type not recognized. Use generic type instead", parentEvent, 4);
|
|
339
|
typeStatus = SpecimenTypeDesignationStatus.TYPE();
|
|
340
|
//TODO use also type info from state
|
|
341
|
}
|
|
342
|
}else{
|
|
343
|
typeStatus = SpecimenTypeDesignationStatus.TYPE();
|
|
344
|
//TODO use also type info from state
|
|
345
|
}
|
|
346
|
|
|
347
|
|
|
348
|
//collection
|
|
349
|
Pattern collectionPattern = Pattern.compile("^" + collectionsRE);
|
|
350
|
matcher = collectionPattern.matcher(singleType);
|
|
351
|
String[] collectionStrings = new String[0];
|
|
352
|
if (matcher.find()){
|
|
353
|
String collectionStr = matcher.group(0);
|
|
354
|
singleType = singleType.substring(collectionStr.length());
|
|
355
|
collectionStr = collectionStr.replace("(", "").replace(")", "").replaceAll("\\s", "");
|
|
356
|
collectionStrings = collectionStr.split(",");
|
|
357
|
}
|
|
358
|
|
|
359
|
//addInfo
|
|
360
|
if (!singleType.isEmpty() && singleType.startsWith(", ")){
|
|
361
|
singleType = singleType.substring(2);
|
|
362
|
}
|
|
363
|
|
|
364
|
boolean notSeen = false;
|
|
365
|
if (singleType.equals("not seen")){
|
|
366
|
singleType = singleType.replace("not seen", "");
|
|
367
|
notSeen = true;
|
|
368
|
}
|
|
369
|
if (singleType.startsWith("not seen, ")){
|
|
370
|
singleType = singleType.replace("not seen, ", "");
|
|
371
|
notSeen = true;
|
|
372
|
}
|
|
373
|
boolean destroyed = false;
|
|
374
|
if (singleType.equals("destroyed")){
|
|
375
|
destroyed = true;
|
|
376
|
singleType = singleType.replace("destroyed", "");
|
|
377
|
}
|
|
378
|
boolean presumedDestroyed = false;
|
|
379
|
if (singleType.equals("presumed destroyed")){
|
|
380
|
presumedDestroyed = true;
|
|
381
|
singleType = singleType.replace("presumed destroyed", "");
|
|
382
|
}
|
|
383
|
boolean hasAddInfo = notSeen || destroyed || presumedDestroyed;
|
|
384
|
|
|
385
|
|
|
386
|
if (!singleType.isEmpty()){
|
|
387
|
String message = "SingleType was not fully read. Remaining: " + singleType + ". Original singleType was: " + singleTypeOrig;
|
|
388
|
fireWarningEvent(message, parentEvent, 6);
|
|
389
|
System.out.println(message);
|
|
390
|
}
|
|
391
|
|
|
392
|
if (collectionStrings.length > 0){
|
|
393
|
boolean isFirst = true;
|
|
394
|
for (String collStr : collectionStrings){
|
|
395
|
Collection collection = getCollection(state, collStr);
|
|
396
|
DerivedUnit unit = isFirst ? facade.innerDerivedUnit()
|
|
397
|
: facade.addDuplicate(collection, null, null, null, null);
|
|
398
|
SpecimenTypeDesignation desig = SpecimenTypeDesignation.NewInstance();
|
|
399
|
designations.add(desig);
|
|
400
|
desig.setTypeSpecimen(unit);
|
|
401
|
desig.setTypeStatus(typeStatus);
|
|
402
|
handleSpecimenTypeAddInfo(state, notSeen, destroyed,
|
|
403
|
presumedDestroyed, desig);
|
|
404
|
name.addTypeDesignation(desig, true);
|
|
405
|
isFirst = false;
|
|
406
|
}
|
|
407
|
}else if (hasAddInfo){ //handle addInfo if no collection data available
|
|
408
|
SpecimenTypeDesignation desig = SpecimenTypeDesignation.NewInstance();
|
|
409
|
designations.add(desig);
|
|
410
|
desig.setTypeStatus(typeStatus);
|
|
411
|
handleSpecimenTypeAddInfo(state, notSeen, destroyed,
|
|
412
|
presumedDestroyed, desig);
|
|
413
|
name.addTypeDesignation(desig, true);
|
|
414
|
}else{
|
|
415
|
fireWarningEvent("No type designation could be created as collection info was not recognized", parentEvent, 4);
|
|
416
|
}
|
|
417
|
}
|
|
418
|
|
|
419
|
if (designatedBy != null){
|
|
420
|
if (designations.size() != 1){
|
|
421
|
fireWarningEvent("Size of type designations is not exactly 1, which is expected for 'designated by'", parentEvent, 2);
|
|
422
|
}
|
|
423
|
designatedBy = designatedBy.trim();
|
|
424
|
if (designatedBy.startsWith("(") && designatedBy.endsWith(")") ){
|
|
425
|
designatedBy = designatedBy.substring(1, designatedBy.length() - 1);
|
|
426
|
}
|
|
427
|
|
|
428
|
for (SpecimenTypeDesignation desig : designations){
|
|
429
|
if (designatedBy.startsWith("designated by")){
|
|
430
|
String titleCache = designatedBy.replace("designated by", "").trim();
|
|
431
|
Reference reference = ReferenceFactory.newGeneric();
|
|
432
|
reference.setTitleCache(titleCache, true);
|
|
433
|
desig.setCitation(reference);
|
|
434
|
//in future we could also try to parse it automatically
|
|
435
|
fireWarningEvent("MANUALLY: Designated by should be parsed manually: " + titleCache, parentEvent, 1);
|
|
436
|
}else if (designatedBy.equals("designated here")){
|
|
437
|
Reference ref = state.getConfig().getSourceReference();
|
|
438
|
desig.setCitation(ref);
|
|
439
|
fireWarningEvent("MANUALLY: Microcitation should be added to 'designated here", parentEvent, 1);
|
|
440
|
}else if (designatedBy.startsWith("according to")){
|
|
441
|
String annotationStr = designatedBy.replace("according to", "").trim();
|
|
442
|
Annotation annotation = Annotation.NewInstance(annotationStr, AnnotationType.EDITORIAL(), Language.ENGLISH());
|
|
443
|
desig.addAnnotation(annotation);
|
|
444
|
}else{
|
|
445
|
fireWarningEvent("Designated by does not match known pattern: " + designatedBy, parentEvent, 6);
|
|
446
|
}
|
|
447
|
}
|
|
448
|
}
|
|
449
|
}else{
|
|
450
|
fireWarningEvent("CollectionAndType unexpectedly not matching: " + collectionAndTypeOrig, parentEvent, 6);
|
|
451
|
}
|
|
452
|
return true;
|
|
453
|
}else{
|
|
454
|
if (state.getConfig().isUseFotGSpecimenTypeCollectionAndTypeOnly()){
|
|
455
|
fireWarningEvent("NO MATCH: " + collectionAndTypeOrig, parentEvent, 4);
|
|
456
|
}
|
|
457
|
return false;
|
|
458
|
}
|
|
459
|
|
|
460
|
// // remove brackets
|
|
461
|
// if (collectionAndType.matches("^\\(.*\\)\\.?$")) {
|
|
462
|
// collectionAndType = collectionAndType.replaceAll("\\.$", "");
|
|
463
|
// collectionAndType = collectionAndType.substring(1, collectionAndType.length() - 1);
|
|
464
|
// }
|
|
465
|
//
|
|
466
|
// String[] split = collectionAndType.split("[;,]");
|
|
467
|
// for (String str : split) {
|
|
468
|
// str = str.trim();
|
|
469
|
// boolean addToAllNamesInGroup = true;
|
|
470
|
// TypeInfo typeInfo = makeSpecimenTypeTypeInfo(str, parentEvent);
|
|
471
|
// SpecimenTypeDesignationStatus typeStatus = typeInfo.status;
|
|
472
|
// Collection collection = this.getCollection(state, typeInfo.collectionString);
|
|
473
|
//
|
|
474
|
// // TODO improve cache strategy handling
|
|
475
|
// DerivedUnit typeSpecimen = facade.addDuplicate(collection, null, null, null, null);
|
|
476
|
// typeSpecimen.setCacheStrategy(new DerivedUnitFacadeCacheStrategy());
|
|
477
|
// name.addSpecimenTypeDesignation(typeSpecimen, typeStatus, null, null, null, false, addToAllNamesInGroup);
|
|
478
|
// }
|
|
479
|
}
|
|
480
|
|
|
481
|
|
|
482
|
/**
|
|
483
|
* @param notSeen
|
|
484
|
* @param destroyed
|
|
485
|
* @param presumedDestroyed
|
|
486
|
* @param desig
|
|
487
|
*/
|
|
488
|
private void handleSpecimenTypeAddInfo(MarkupImportState state, boolean notSeen, boolean destroyed,
|
|
489
|
boolean presumedDestroyed, SpecimenTypeDesignation desig) {
|
|
490
|
DerivedUnit specimen = desig.getTypeSpecimen();
|
|
491
|
AnnotatableEntity annotEntity = specimen != null ? specimen : desig;
|
|
492
|
|
|
493
|
if (notSeen){
|
|
494
|
UUID uuidNotSeenMarker = MarkupTransformer.uuidMarkerNotSeen;
|
|
495
|
MarkerType notSeenMarkerType = getMarkerType(state, uuidNotSeenMarker, "Not seen", "Not seen", null, null);
|
|
496
|
Marker marker = Marker.NewInstance(notSeenMarkerType, true);
|
|
497
|
annotEntity.addMarker(marker);
|
|
498
|
fireWarningEvent("not seen not yet implemented", "handleSpecimenTypeAddInfo", 4);
|
|
499
|
}
|
|
500
|
if (destroyed){
|
|
501
|
UUID uuidDestroyedMarker = MarkupTransformer.uuidMarkerDestroyed;
|
|
502
|
MarkerType destroyedMarkerType = getMarkerType(state, uuidDestroyedMarker, "Destroyed", "Destroyed", null, null);
|
|
503
|
Marker marker = Marker.NewInstance(destroyedMarkerType, true);
|
|
504
|
annotEntity.addMarker(marker);
|
|
505
|
fireWarningEvent("'destroyed' not yet fully implemented", "handleSpecimenTypeAddInfo", 4);
|
|
506
|
}
|
|
507
|
if (presumedDestroyed){
|
|
508
|
Annotation annotation = Annotation.NewInstance("presumably destroyed", Language.ENGLISH());
|
|
509
|
annotation.setAnnotationType(AnnotationType.EDITORIAL());
|
|
510
|
annotEntity.addAnnotation(annotation);
|
|
511
|
}
|
|
512
|
}
|
|
513
|
|
|
514
|
|
|
515
|
private TypeInfo makeSpecimenTypeTypeInfo(MarkupImportState state, String originalString, TypeInfo lastTypeInfo, XMLEvent event) {
|
|
516
|
TypeInfo result = new TypeInfo();
|
|
517
|
if ("not designated".equals(originalString)){
|
|
518
|
result.notDesignated = true;
|
|
519
|
return result;
|
|
520
|
}
|
|
521
|
List<String> knownCollections = state.getConfig().getKnownCollections();
|
|
522
|
for (String knownCollection:knownCollections){
|
|
523
|
if (originalString.contains(knownCollection)){
|
|
524
|
result.collectionString = knownCollection;
|
|
525
|
originalString = originalString.replace(knownCollection, "").trim();
|
|
526
|
break;
|
|
527
|
}
|
|
528
|
}
|
|
529
|
|
|
530
|
String[] split = originalString.split("(?<!not)\\s+");
|
|
531
|
|
|
532
|
String unrecognizedTypeParts = null;
|
|
533
|
for (String str : split) {
|
|
534
|
//holo/lecto/iso ...
|
|
535
|
if (str.matches(SpecimenTypeParser.typeTypePattern)) {
|
|
536
|
SpecimenTypeDesignationStatus status;
|
|
537
|
try {
|
|
538
|
status = SpecimenTypeParser.parseSpecimenTypeStatus(str);
|
|
539
|
} catch (UnknownCdmTypeException e) {
|
|
540
|
String message = "Specimen type status '%s' not recognized by parser";
|
|
541
|
fireWarningEvent(String.format(message, str), event, 4);
|
|
542
|
status = null;
|
|
543
|
}
|
|
544
|
if (result.status != null){
|
|
545
|
String message = "More than 1 status string found: " + originalString;
|
|
546
|
fireWarningEvent(message, event, 4);
|
|
547
|
}
|
|
548
|
result.status = status;
|
|
549
|
} else if (str.matches(SpecimenTypeParser.collectionPattern)) {
|
|
550
|
if (result.collectionString != null){
|
|
551
|
String message = "More than 1 collection string found: " + originalString;
|
|
552
|
fireWarningEvent(message, event, 4);
|
|
553
|
}
|
|
554
|
result.collectionString = str;
|
|
555
|
} else if (str.matches(SpecimenTypeParser.notSeen)) {
|
|
556
|
if (result.notSeen){
|
|
557
|
String message = "More than 1 'not seen' string found: " + originalString;
|
|
558
|
fireWarningEvent(message, event, 4);
|
|
559
|
}
|
|
560
|
result.notSeen = true;
|
|
561
|
} else {
|
|
562
|
unrecognizedTypeParts = CdmUtils.concat(" ", unrecognizedTypeParts, str);
|
|
563
|
}
|
|
564
|
if (result.status == null && lastTypeInfo != null && lastTypeInfo.status != null){
|
|
565
|
result.status = lastTypeInfo.status;
|
|
566
|
}
|
|
567
|
}
|
|
568
|
if(isNotBlank(unrecognizedTypeParts)){
|
|
569
|
String message = "Type parts '%s' could not be recognized";
|
|
570
|
fireWarningEvent(String.format(message, unrecognizedTypeParts), event, 2);
|
|
571
|
}
|
|
572
|
return result;
|
|
573
|
}
|
|
574
|
|
|
575
|
|
|
576
|
private void handleGathering(MarkupImportState state, XMLEventReader readerOrig, XMLEvent parentEvent , DerivedUnitFacade facade) throws XMLStreamException {
|
|
577
|
checkNoAttributes(parentEvent);
|
|
578
|
boolean hasCollector = false;
|
|
579
|
boolean hasFieldNum = false;
|
|
580
|
|
|
581
|
LookAheadEventReader reader = new LookAheadEventReader(parentEvent.asStartElement(), readerOrig);
|
|
582
|
|
|
583
|
// elements
|
|
584
|
while (reader.hasNext()) {
|
|
585
|
XMLEvent next = readNoWhitespace(reader);
|
|
586
|
if (isMyEndingElement(next, parentEvent)) {
|
|
587
|
if (! hasCollector){
|
|
588
|
if (state.getCurrentCollector() == null){
|
|
589
|
checkMandatoryElement(hasCollector,parentEvent.asStartElement(), COLLECTOR);
|
|
590
|
}else{
|
|
591
|
facade.setCollector(state.getCurrentCollector());
|
|
592
|
}
|
|
593
|
}
|
|
594
|
checkMandatoryElement(hasFieldNum,parentEvent.asStartElement(), FIELD_NUM);
|
|
595
|
return;
|
|
596
|
}else if (isStartingElement(next, COLLECTOR)) {
|
|
597
|
hasCollector = true;
|
|
598
|
String collectorStr = getCData(state, reader, next);
|
|
599
|
TeamOrPersonBase<?> collector = createCollector(state, collectorStr);
|
|
600
|
facade.setCollector(collector);
|
|
601
|
state.setCurrentCollector(collector);
|
|
602
|
} else if (isStartingElement(next, ALTERNATIVE_COLLECTOR)) {
|
|
603
|
handleNotYetImplementedElement(next);
|
|
604
|
} else if (isStartingElement(next, FIELD_NUM)) {
|
|
605
|
hasFieldNum = true;
|
|
606
|
String fieldNumStr = getCData(state, reader, next);
|
|
607
|
facade.setFieldNumber(fieldNumStr);
|
|
608
|
} else if (isStartingElement(next, ALTERNATIVE_FIELD_NUM)) {
|
|
609
|
handleAlternativeFieldNumber(state, reader, next, facade.innerFieldUnit());
|
|
610
|
} else if (isStartingElement(next, COLLECTION_TYPE_STATUS)) {
|
|
611
|
handleNotYetImplementedElement(next);
|
|
612
|
} else if (isStartingElement(next, COLLECTION_AND_TYPE)) {
|
|
613
|
handleGatheringCollectionAndType(state, reader, next, facade);
|
|
614
|
} else if (isStartingElement(next, ALTERNATIVE_COLLECTION_TYPE_STATUS)) {
|
|
615
|
handleNotYetImplementedElement(next);
|
|
616
|
} else if (isStartingElement(next, SUB_GATHERING)) {
|
|
617
|
handleNotYetImplementedElement(next);
|
|
618
|
} else if (isStartingElement(next, COLLECTION)) {
|
|
619
|
handleNotYetImplementedElement(next);
|
|
620
|
} else if (isStartingElement(next, LOCALITY)) {
|
|
621
|
handleLocality(state, reader, next, facade);
|
|
622
|
} else if (isStartingElement(next, FULL_NAME)) {
|
|
623
|
Rank defaultRank = Rank.SPECIES(); // can be any
|
|
624
|
INonViralName nvn = createNameByCode(state, defaultRank);
|
|
625
|
handleFullName(state, reader, nvn, next);
|
|
626
|
TaxonName name = TaxonName.castAndDeproxy(nvn);
|
|
627
|
DeterminationEvent.NewInstance(name, facade.innerDerivedUnit() != null ? facade.innerDerivedUnit() : facade.innerFieldUnit());
|
|
628
|
} else if (isStartingElement(next, DATES)) {
|
|
629
|
TimePeriod timePeriod = handleDates(state, reader, next);
|
|
630
|
facade.setGatheringPeriod(timePeriod);
|
|
631
|
} else if (isStartingElement(next, GATHERING_NOTES)) {
|
|
632
|
handleAmbigousManually(state, reader, next.asStartElement());
|
|
633
|
} else if (isStartingElement(next, NOTES)) {
|
|
634
|
handleNotYetImplementedElement(next);
|
|
635
|
}else if (next.isCharacters()) {
|
|
636
|
String text = next.asCharacters().getData().trim();
|
|
637
|
if (isPunctuation(text)){
|
|
638
|
//do nothing
|
|
639
|
}else if (state.isSpecimenType() && charIsSimpleType(text) ){
|
|
640
|
//do nothing
|
|
641
|
}else if ( (text.equals("=") || text.equals("(") ) && reader.nextIsStart(ALTERNATIVE_FIELD_NUM)){
|
|
642
|
//do nothing
|
|
643
|
}else if ( (text.equals(").") || text.equals(")")) && reader.previousWasEnd(ALTERNATIVE_FIELD_NUM)){
|
|
644
|
//do nothing
|
|
645
|
}else if ( charIsOpeningOrClosingBracket(text) ){
|
|
646
|
//for now we don't do anything, however in future brackets may have semantics
|
|
647
|
}else{
|
|
648
|
//TODO
|
|
649
|
String message = "Unrecognized text: %s";
|
|
650
|
fireWarningEvent(String.format(message, text), next, 6);
|
|
651
|
}
|
|
652
|
} else {
|
|
653
|
handleUnexpectedElement(next);
|
|
654
|
}
|
|
655
|
}
|
|
656
|
throw new IllegalStateException("Collection has no closing tag.");
|
|
657
|
|
|
658
|
}
|
|
659
|
|
|
660
|
|
|
661
|
private final String fotgPattern = "^\\(([A-Z]{1,3})(?:,\\s?([A-Z]{1,3}))*\\)"; // eg. (US, B, CAN)
|
|
662
|
private void handleGatheringCollectionAndType(MarkupImportState state, XMLEventReader reader, XMLEvent parent, DerivedUnitFacade facade) throws XMLStreamException {
|
|
663
|
checkNoAttributes(parent);
|
|
664
|
|
|
665
|
XMLEvent next = readNoWhitespace(reader);
|
|
666
|
|
|
667
|
if (next.isCharacters()){
|
|
668
|
String txt = next.asCharacters().getData().trim();
|
|
669
|
if (state.isSpecimenType()){
|
|
670
|
state.addCollectionAndType(txt);
|
|
671
|
}else{
|
|
672
|
|
|
673
|
Matcher fotgMatcher = Pattern.compile(fotgPattern).matcher(txt);
|
|
674
|
|
|
675
|
if (fotgMatcher.matches()){
|
|
676
|
txt = txt.substring(1, txt.length() - 1); //remove bracket
|
|
677
|
String[] splits = txt.split(",");
|
|
678
|
for (String split : splits ){
|
|
679
|
Collection collection = getCollection(state, split.trim());
|
|
680
|
if (facade.innerDerivedUnit() == null){
|
|
681
|
String message = "Adding a duplicate to a non derived unit based facade is not possible. Please check why no derived unit exists yet in facade!";
|
|
682
|
this.fireWarningEvent(message, next, -6);
|
|
683
|
}else{
|
|
684
|
facade.addDuplicate(collection, null, null, null, null);
|
|
685
|
}
|
|
686
|
}
|
|
687
|
//FIXME 9
|
|
688
|
//create derived units and and add collections
|
|
689
|
|
|
690
|
}else{
|
|
691
|
fireWarningEvent("Collection and type pattern for gathering not recognized: " + txt, next, 4);
|
|
692
|
}
|
|
693
|
}
|
|
694
|
|
|
695
|
}else{
|
|
696
|
fireUnexpectedEvent(next, 0);
|
|
697
|
}
|
|
698
|
|
|
699
|
if (isMyEndingElement(next, parent)){
|
|
700
|
return; //in case we have a completely empty element
|
|
701
|
}
|
|
702
|
next = readNoWhitespace(reader);
|
|
703
|
if (isMyEndingElement(next, parent)){
|
|
704
|
return;
|
|
705
|
}else{
|
|
706
|
fireUnexpectedEvent(next, 0);
|
|
707
|
return;
|
|
708
|
}
|
|
709
|
}
|
|
710
|
|
|
711
|
|
|
712
|
private Collection getCollection(MarkupImportState state, String code) {
|
|
713
|
Collection collection = state.getCollectionByCode(code);
|
|
714
|
if (collection == null){
|
|
715
|
List<Collection> list = this.docImport.getCollectionService().searchByCode(code);
|
|
716
|
if (list.size() == 1){
|
|
717
|
collection = list.get(0);
|
|
718
|
}else if (list.size() > 1){
|
|
719
|
fireWarningEvent("More then one occurrence for collection " + code + " in database. Collection not reused" , "", 1);
|
|
720
|
}
|
|
721
|
|
|
722
|
if (collection == null){
|
|
723
|
collection = Collection.NewInstance();
|
|
724
|
collection.setCode(code);
|
|
725
|
this.docImport.getCollectionService().saveOrUpdate(collection);
|
|
726
|
}
|
|
727
|
state.putCollectionByCode(code, collection);
|
|
728
|
}
|
|
729
|
return collection;
|
|
730
|
}
|
|
731
|
|
|
732
|
|
|
733
|
private void handleAlternativeFieldNumber(MarkupImportState state, XMLEventReader reader, XMLEvent parent, FieldUnit fieldUnit) throws XMLStreamException {
|
|
734
|
Map<String, Attribute> attrs = getAttributes(parent);
|
|
735
|
Boolean doubtful = this.getAndRemoveBooleanAttributeValue(parent, attrs, "doubful", false);
|
|
736
|
|
|
737
|
//for now we do not handle annotation and typeNotes
|
|
738
|
String altFieldNum = getCData(state, reader, parent, false).trim();
|
|
739
|
DefinedTerm type = this.getIdentifierType(state, MarkupTransformer.uuidIdentTypeAlternativeFieldNumber, "Alternative field number", "Alternative field number", "alt. field no.", null);
|
|
740
|
fieldUnit.addIdentifier(altFieldNum, type);
|
|
741
|
if (doubtful){
|
|
742
|
fireWarningEvent("Marking alternative field numbers as doubtful not yet possible, see #4673", parent,4);
|
|
743
|
// Marker.NewInstance(identifier, "true", MarkerType.IS_DOUBTFUL());
|
|
744
|
}
|
|
745
|
|
|
746
|
}
|
|
747
|
|
|
748
|
|
|
749
|
private boolean charIsOpeningOrClosingBracket(String text) {
|
|
750
|
return text.equals("(") || text.equals(")");
|
|
751
|
}
|
|
752
|
|
|
753
|
|
|
754
|
private TimePeriod handleDates(MarkupImportState state, XMLEventReader reader, XMLEvent parent) throws XMLStreamException {
|
|
755
|
checkNoAttributes(parent);
|
|
756
|
TimePeriod result = TimePeriod.NewInstance();
|
|
757
|
String parseMessage = "%s can not be parsed: %s";
|
|
758
|
boolean hasFullDate = false;
|
|
759
|
boolean hasAtomised = false;
|
|
760
|
boolean hasUnparsedAtomised = false;
|
|
761
|
while (reader.hasNext()) {
|
|
762
|
XMLEvent next = readNoWhitespace(reader);
|
|
763
|
if (isMyEndingElement(next, parent)) {
|
|
764
|
if (! isAlternative(hasFullDate, hasAtomised, hasUnparsedAtomised)){
|
|
765
|
String message = "Some problems exist when defining the date";
|
|
766
|
fireWarningEvent(message, parent, 4);
|
|
767
|
}
|
|
768
|
return result;
|
|
769
|
} else if (isStartingElement(next, FULL_DATE)) {
|
|
770
|
String fullDate = getCData(state, reader, next, true);
|
|
771
|
if (fullDate.endsWith(".")){
|
|
772
|
fullDate = fullDate.substring(0, fullDate.length()-1);
|
|
773
|
}
|
|
774
|
result = TimePeriodParser.parseString(fullDate);
|
|
775
|
if (result.getFreeText() != null){
|
|
776
|
fireWarningEvent(String.format(parseMessage, FULL_DATE, fullDate), parent, 1);
|
|
777
|
}
|
|
778
|
hasFullDate = true;
|
|
779
|
} else if (isStartingElement(next, DAY)) {
|
|
780
|
String day = getCData(state, reader, next, true).trim();
|
|
781
|
day = normalizeDate(day);
|
|
782
|
if (CdmUtils.isNumeric(day)){
|
|
783
|
result.setStartDay(Integer.valueOf(day));
|
|
784
|
hasAtomised = true;
|
|
785
|
}else{
|
|
786
|
fireWarningEvent(String.format(parseMessage,"Day", day), parent, 2);
|
|
787
|
hasUnparsedAtomised = true;
|
|
788
|
}
|
|
789
|
} else if (isStartingElement(next, MONTH)) {
|
|
790
|
String month = getCData(state, reader, next, true).trim();
|
|
791
|
month = normalizeDate(month);
|
|
792
|
if (CdmUtils.isNumeric(month)){
|
|
793
|
result.setStartMonth(Integer.valueOf(month));
|
|
794
|
hasAtomised = true;
|
|
795
|
}else{
|
|
796
|
fireWarningEvent(String.format(parseMessage,"Month", month), parent, 2);
|
|
797
|
hasUnparsedAtomised = true;
|
|
798
|
}
|
|
799
|
} else if (isStartingElement(next, YEAR)) {
|
|
800
|
String year = getCData(state, reader, next, true).trim();
|
|
801
|
year = normalizeDate(year);
|
|
802
|
if (CdmUtils.isNumeric(year)){
|
|
803
|
result.setStartYear(Integer.valueOf(year));
|
|
804
|
hasAtomised = true;
|
|
805
|
}else{
|
|
806
|
fireWarningEvent(String.format(parseMessage,"Year", year), parent, 2);
|
|
807
|
hasUnparsedAtomised = true;
|
|
808
|
}
|
|
809
|
} else {
|
|
810
|
handleUnexpectedElement(next);
|
|
811
|
}
|
|
812
|
}
|
|
813
|
throw new IllegalStateException("Dates has no closing tag.");
|
|
814
|
}
|
|
815
|
|
|
816
|
|
|
817
|
private String normalizeDate(String partOfDate) {
|
|
818
|
if (isBlank(partOfDate)){
|
|
819
|
return null;
|
|
820
|
}
|
|
821
|
partOfDate = partOfDate.trim();
|
|
822
|
while (partOfDate.startsWith("-")){
|
|
823
|
partOfDate = partOfDate.substring(1);
|
|
824
|
}
|
|
825
|
return partOfDate;
|
|
826
|
}
|
|
827
|
|
|
828
|
|
|
829
|
private boolean isAlternative(boolean first, boolean second, boolean third) {
|
|
830
|
return ( (first ^ second) && !third) ||
|
|
831
|
(! first && ! second && third) ;
|
|
832
|
}
|
|
833
|
|
|
834
|
|
|
835
|
private void handleLocality(MarkupImportState state, XMLEventReader reader,XMLEvent parentEvent, DerivedUnitFacade facade)throws XMLStreamException {
|
|
836
|
String classValue = getClassOnlyAttribute(parentEvent);
|
|
837
|
boolean isLocality = false;
|
|
838
|
NamedAreaLevel areaLevel = null;
|
|
839
|
if ("locality".equalsIgnoreCase(classValue)||state.getConfig().isIgnoreLocalityClass()) {
|
|
840
|
isLocality = true;
|
|
841
|
} else {
|
|
842
|
areaLevel = makeNamedAreaLevel(state, classValue, parentEvent);
|
|
843
|
}
|
|
844
|
|
|
845
|
String text = "";
|
|
846
|
// elements
|
|
847
|
while (reader.hasNext()) {
|
|
848
|
XMLEvent next = readNoWhitespace(reader);
|
|
849
|
if (isMyEndingElement(next, parentEvent)) {
|
|
850
|
if (StringUtils.isNotBlank(text)) {
|
|
851
|
text = normalize(text);text = removeTrailingPunctuation(text);
|
|
852
|
if (isLocality) {
|
|
853
|
facade.setLocality(text, getDefaultLanguage(state));
|
|
854
|
} else {
|
|
855
|
text = CdmUtils.removeTrailingDots(text);
|
|
856
|
NamedArea area = makeArea(state, text, areaLevel);
|
|
857
|
facade.addCollectingArea(area);
|
|
858
|
}
|
|
859
|
}
|
|
860
|
// TODO
|
|
861
|
return;
|
|
862
|
}else if (isStartingElement(next, ALTITUDE)) {
|
|
863
|
handleNotYetImplementedElement(next);
|
|
864
|
// homotypicalGroup = handleNom(state, reader, next, taxon,
|
|
865
|
// homotypicalGroup);
|
|
866
|
} else if (isStartingElement(next, COORDINATES)) {
|
|
867
|
handleNotYetImplementedElement(next);
|
|
868
|
} else if (isStartingElement(next, ANNOTATION)) {
|
|
869
|
handleNotYetImplementedElement(next);
|
|
870
|
} else if (next.isCharacters()) {
|
|
871
|
text += next.asCharacters().getData();
|
|
872
|
} else {
|
|
873
|
handleUnexpectedElement(next);
|
|
874
|
}
|
|
875
|
}
|
|
876
|
throw new IllegalStateException("<SpecimenType> has no closing tag");
|
|
877
|
}
|
|
878
|
|
|
879
|
|
|
880
|
|
|
881
|
/**
|
|
882
|
* @param text
|
|
883
|
* @return
|
|
884
|
*/
|
|
885
|
private String removeTrailingPunctuation(String text) {
|
|
886
|
while (isPunctuation(text.substring(text.length()-1))){
|
|
887
|
text = text.substring(0, text.length()-1).trim();
|
|
888
|
}
|
|
889
|
return text;
|
|
890
|
}
|
|
891
|
|
|
892
|
|
|
893
|
private TeamOrPersonBase<?> createCollector(MarkupImportState state, String collectorStr) {
|
|
894
|
return createAuthor(state, collectorStr);
|
|
895
|
}
|
|
896
|
|
|
897
|
|
|
898
|
public List<DescriptionElementBase> handleMaterialsExamined(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Feature feature, TaxonDescription defaultDescription) throws XMLStreamException {
|
|
899
|
List<DescriptionElementBase> result = new ArrayList<DescriptionElementBase>();
|
|
900
|
//reset current areas
|
|
901
|
state.removeCurrentAreas();
|
|
902
|
while (reader.hasNext()) {
|
|
903
|
XMLEvent next = readNoWhitespace(reader);
|
|
904
|
if (isMyEndingElement(next, parentEvent)) {
|
|
905
|
if (result.isEmpty()){
|
|
906
|
fireWarningEvent("Materials examined created empty Individual Associations list", parentEvent, 4);
|
|
907
|
}
|
|
908
|
state.removeCurrentAreas();
|
|
909
|
return result;
|
|
910
|
} else if (isStartingElement(next, SUB_HEADING)) {
|
|
911
|
// Map<String, Object> inlineMarkup = new HashMap<String, Object>();
|
|
912
|
String text = getCData(state, reader, next, true);
|
|
913
|
if (isFeatureHeading(state, next, text)){
|
|
914
|
feature = makeHeadingFeature(state, next, text, feature);
|
|
915
|
}else{
|
|
916
|
String message = "Unhandled subheading: %s";
|
|
917
|
fireWarningEvent(String.format(message, text), next, 4);
|
|
918
|
}
|
|
919
|
// for (String key : inlineMarkup.keySet()){
|
|
920
|
// handleInlineMarkup(state, key, inlineMarkup);
|
|
921
|
// }
|
|
922
|
|
|
923
|
} else if (isStartingElement(next, BR) || isEndingElement(next, BR)) {
|
|
924
|
//do nothing
|
|
925
|
} else if (isStartingElement(next, GATHERING)) {
|
|
926
|
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(SpecimenOrObservationType.DerivedUnit);
|
|
927
|
addCurrentAreas(state, next, facade);
|
|
928
|
handleGathering(state, reader, next, facade);
|
|
929
|
SpecimenOrObservationBase<?> specimen;
|
|
930
|
if (facade.innerDerivedUnit() != null){
|
|
931
|
specimen = facade.innerDerivedUnit();
|
|
932
|
}else{
|
|
933
|
specimen = facade.innerFieldUnit();
|
|
934
|
}
|
|
935
|
IndividualsAssociation individualsAssociation = IndividualsAssociation.NewInstance();
|
|
936
|
individualsAssociation.addPrimaryTaxonomicSource(state.getConfig().getSourceReference());
|
|
937
|
individualsAssociation.setAssociatedSpecimenOrObservation(specimen);
|
|
938
|
result.add(individualsAssociation);
|
|
939
|
} else if (isStartingElement(next, GATHERING_GROUP)) {
|
|
940
|
List<DescriptionElementBase> list = getGatheringGroupDescription(state, reader, next);
|
|
941
|
result.addAll(list);
|
|
942
|
}else if (next.isCharacters()) {
|
|
943
|
String text = next.asCharacters().getData().trim();
|
|
944
|
if (isPunctuation(text)){
|
|
945
|
//do nothing
|
|
946
|
}else{
|
|
947
|
String message = "Unrecognized text: %s";
|
|
948
|
fireWarningEvent(String.format(message, text), next, 6);
|
|
949
|
}
|
|
950
|
} else {
|
|
951
|
handleUnexpectedElement(next);
|
|
952
|
}
|
|
953
|
}
|
|
954
|
throw new IllegalStateException("<String> has no closing tag");
|
|
955
|
|
|
956
|
}
|
|
957
|
|
|
958
|
|
|
959
|
|
|
960
|
private List<DescriptionElementBase> getGatheringGroupDescription(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
961
|
Map<String, Attribute> attributes = getAttributes(parentEvent);
|
|
962
|
String geoScope = getAndRemoveAttributeValue(attributes, "geoscope");
|
|
963
|
Boolean doubtful = getAndRemoveBooleanAttributeValue(parentEvent, attributes, DOUBTFUL, null);
|
|
964
|
checkNoAttributes(attributes, parentEvent);
|
|
965
|
|
|
966
|
List<DescriptionElementBase> result = new ArrayList<DescriptionElementBase>();
|
|
967
|
|
|
968
|
|
|
969
|
TaxonDescription td = null;
|
|
970
|
|
|
971
|
if (isNotBlank(geoScope)){
|
|
972
|
NamedArea area = Country.getCountryByLabel(geoScope);
|
|
973
|
if (area == null){
|
|
974
|
try {
|
|
975
|
area = state.getTransformer().getNamedAreaByKey(geoScope);
|
|
976
|
} catch (Exception e) {
|
|
977
|
fireWarningEvent("getNamedArea not supported", parentEvent, 16);
|
|
978
|
}
|
|
979
|
}
|
|
980
|
if (area == null){
|
|
981
|
fireWarningEvent("Area for geoscope not found: " + geoScope +"; add specimen group to ordinary description", parentEvent, 4);
|
|
982
|
}else{
|
|
983
|
state.addCurrentArea(area);
|
|
984
|
Set<TaxonDescription> descs = state.getCurrentTaxon().getDescriptions();
|
|
985
|
for (TaxonDescription desc : descs){
|
|
986
|
Set<NamedArea> scopes = desc.getGeoScopes();
|
|
987
|
if (scopes.size() == 1 && scopes.iterator().next().equals(area)){
|
|
988
|
td = desc;
|
|
989
|
break;
|
|
990
|
}
|
|
991
|
}
|
|
992
|
if (td == null){
|
|
993
|
TaxonDescription desc = TaxonDescription.NewInstance(state.getCurrentTaxon());
|
|
994
|
desc.addPrimaryTaxonomicSource(state.getConfig().getSourceReference(), null);
|
|
995
|
desc.addGeoScope(area);
|
|
996
|
if (doubtful != null){
|
|
997
|
desc.addMarker(Marker.NewInstance(MarkerType.IS_DOUBTFUL(), doubtful));
|
|
998
|
}
|
|
999
|
td = desc;
|
|
1000
|
}
|
|
1001
|
}
|
|
1002
|
}
|
|
1003
|
|
|
1004
|
while (reader.hasNext()) {
|
|
1005
|
XMLEvent next = readNoWhitespace(reader);
|
|
1006
|
if (isMyEndingElement(next, parentEvent)) {
|
|
1007
|
if (result.isEmpty()){
|
|
1008
|
fireWarningEvent("Gathering group created empty Individual Associations list", parentEvent, 4);
|
|
1009
|
}
|
|
1010
|
state.removeCurrentAreas();
|
|
1011
|
return result;
|
|
1012
|
} else if (isStartingElement(next, GATHERING)) {
|
|
1013
|
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(SpecimenOrObservationType.DerivedUnit);
|
|
1014
|
addCurrentAreas(state, next, facade);
|
|
1015
|
handleGathering(state, reader, next, facade);
|
|
1016
|
SpecimenOrObservationBase<?> specimen;
|
|
1017
|
if (facade.innerDerivedUnit() != null){
|
|
1018
|
specimen = facade.innerDerivedUnit();
|
|
1019
|
}else{
|
|
1020
|
specimen = facade.innerFieldUnit();
|
|
1021
|
}
|
|
1022
|
IndividualsAssociation individualsAssociation = IndividualsAssociation.NewInstance();
|
|
1023
|
individualsAssociation.addPrimaryTaxonomicSource(state.getConfig().getSourceReference());
|
|
1024
|
individualsAssociation.setAssociatedSpecimenOrObservation(specimen);
|
|
1025
|
result.add(individualsAssociation);
|
|
1026
|
|
|
1027
|
}else if (next.isCharacters()) {
|
|
1028
|
String text = next.asCharacters().getData().trim();
|
|
1029
|
if (isPunctuation(text)){
|
|
1030
|
//do nothing
|
|
1031
|
}else{
|
|
1032
|
//TODO
|
|
1033
|
String message = "Unrecognized text: %s";
|
|
1034
|
fireWarningEvent(String.format(message, text), next, 6);
|
|
1035
|
}
|
|
1036
|
} else {
|
|
1037
|
handleUnexpectedElement(next);
|
|
1038
|
}
|
|
1039
|
}
|
|
1040
|
throw new IllegalStateException("<Gathering group> has no closing tag");
|
|
1041
|
|
|
1042
|
}
|
|
1043
|
|
|
1044
|
private void addCurrentAreas(MarkupImportState state, XMLEvent event, DerivedUnitFacade facade) {
|
|
1045
|
for (NamedArea area : state.getCurrentAreas()){
|
|
1046
|
if (area == null){
|
|
1047
|
continue;
|
|
1048
|
}else if (area.isInstanceOf(Country.class)){
|
|
1049
|
facade.setCountry(area);
|
|
1050
|
}else{
|
|
1051
|
String message = "Current area %s is not country. This is not expected for currently known data.";
|
|
1052
|
fireWarningEvent(String.format(message, area.getTitleCache()), event, 2);
|
|
1053
|
facade.addCollectingArea(area);
|
|
1054
|
}
|
|
1055
|
}
|
|
1056
|
|
|
1057
|
}
|
|
1058
|
|
|
1059
|
|
|
1060
|
// private void handleInlineMarkup(MarkupImportState state, String key, Map<String, Object> inlineMarkup) {
|
|
1061
|
// Object obj = inlineMarkup.get(key);
|
|
1062
|
// if (key.equals(LOCALITY)){
|
|
1063
|
// if (obj instanceof NamedArea){
|
|
1064
|
// NamedArea area = (NamedArea)obj;
|
|
1065
|
// state.addCurrentArea(area);
|
|
1066
|
// }
|
|
1067
|
// }
|
|
1068
|
//
|
|
1069
|
// }
|
|
1070
|
|
|
1071
|
|
|
1072
|
/**
|
|
1073
|
* Changes the feature if the (sub)-heading implies this. Also recognizes hidden country information
|
|
1074
|
* @param state
|
|
1075
|
* @param parent
|
|
1076
|
* @param text
|
|
1077
|
* @param feature
|
|
1078
|
* @return
|
|
1079
|
*/
|
|
1080
|
private Feature makeHeadingFeature(MarkupImportState state, XMLEvent parent, String originalText, Feature feature) {
|
|
1081
|
//expand, provide by config or service
|
|
1082
|
String materialRegEx = "Mat[\u00E9\u00C9]riel";
|
|
1083
|
String examinedRegEx = "[\u00E9\u00C9]tudi[\u00E9\u00C9]";
|
|
1084
|
String countryRegEx = "(gabonais)";
|
|
1085
|
String postfixCountryRegEx = "\\s+(pour le Gabon)";
|
|
1086
|
|
|
1087
|
String materialExaminedRegEx = "(?i)" + materialRegEx + "\\s+(" + countryRegEx +"\\s+)?" + examinedRegEx + "(" +postfixCountryRegEx + ")?:?";
|
|
1088
|
|
|
1089
|
String text = originalText;
|
|
1090
|
|
|
1091
|
if (isBlank(text)){
|
|
1092
|
return feature;
|
|
1093
|
}else{
|
|
1094
|
if (text.matches(materialExaminedRegEx)){
|
|
1095
|
//gabon specific
|
|
1096
|
if (text.contains("gabonais ")){
|
|
1097
|
text = text.replace("gabonais ", "");
|
|
1098
|
state.addCurrentArea(Country.GABONGABONESEREPUBLIC());
|
|
1099
|
}
|
|
1100
|
if (text.contains(" pour le Gabon")){
|
|
1101
|
text = text.replace(" pour le Gabon", "");
|
|
1102
|
state.addCurrentArea(Country.GABONGABONESEREPUBLIC());
|
|
1103
|
}
|
|
1104
|
|
|
1105
|
//update feature
|
|
1106
|
feature = Feature.MATERIALS_EXAMINED();
|
|
1107
|
state.putFeatureToGeneralSorterList(feature);
|
|
1108
|
return feature;
|
|
1109
|
}else{
|
|
1110
|
String message = "Heading/Subheading not recognized: %s";
|
|
1111
|
fireWarningEvent(String.format(message, originalText), parent, 4);
|
|
1112
|
return feature;
|
|
1113
|
}
|
|
1114
|
}
|
|
1115
|
}
|
|
1116
|
|
|
1117
|
|
|
1118
|
/**
|
|
1119
|
* True if heading or subheading represents feature information
|
|
1120
|
* @param state
|
|
1121
|
* @param parent
|
|
1122
|
* @param text
|
|
1123
|
* @return
|
|
1124
|
*/
|
|
1125
|
private boolean isFeatureHeading(MarkupImportState state, XMLEvent parent, String text) {
|
|
1126
|
return makeHeadingFeature(state, parent, text, null) != null;
|
|
1127
|
}
|
|
1128
|
|
|
1129
|
|
|
1130
|
public String handleInLineGathering(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
1131
|
DerivedUnitFacade facade = DerivedUnitFacade.NewInstance(SpecimenOrObservationType.FieldUnit);
|
|
1132
|
handleGathering(state, reader, parentEvent, facade);
|
|
1133
|
SpecimenOrObservationBase<?> specimen = facade.innerFieldUnit();
|
|
1134
|
if (specimen == null){
|
|
1135
|
specimen = facade.innerDerivedUnit();
|
|
1136
|
String message = "Inline gaterhing has no field unit";
|
|
1137
|
fireWarningEvent(message, parentEvent, 2);
|
|
1138
|
}
|
|
1139
|
|
|
1140
|
String result = "<cdm:specimen uuid='%s'>%s</specimen>";
|
|
1141
|
if (specimen != null){
|
|
1142
|
result = String.format(result, specimen.getUuid(), specimen.getTitleCache());
|
|
1143
|
}else{
|
|
1144
|
String message = "Inline gathering has no specimen";
|
|
1145
|
fireWarningEvent(message, parentEvent, 4);
|
|
1146
|
}
|
|
1147
|
save(specimen, state);
|
|
1148
|
return result;
|
|
1149
|
}
|
|
1150
|
|
|
1151
|
|
|
1152
|
|
|
1153
|
|
|
1154
|
|
|
1155
|
}
|