|
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
|
package eu.etaxonomy.cdm.io.tcsrdf;
|
|
10
|
|
|
11
|
import java.util.ArrayList;
|
|
12
|
import java.util.Iterator;
|
|
13
|
import java.util.List;
|
|
14
|
import java.util.Set;
|
|
15
|
|
|
16
|
import org.apache.log4j.Logger;
|
|
17
|
import org.jdom.Attribute;
|
|
18
|
import org.jdom.Element;
|
|
19
|
import org.jdom.Namespace;
|
|
20
|
import org.jdom.filter.ElementFilter;
|
|
21
|
import org.jdom.filter.Filter;
|
|
22
|
import org.springframework.stereotype.Component;
|
|
23
|
|
|
24
|
import eu.etaxonomy.cdm.api.service.ITaxonService;
|
|
25
|
import eu.etaxonomy.cdm.common.XmlHelp;
|
|
26
|
import eu.etaxonomy.cdm.io.common.ICdmIO;
|
|
27
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
|
28
|
import eu.etaxonomy.cdm.io.common.MapWrapper;
|
|
29
|
import eu.etaxonomy.cdm.model.common.DescriptionElementSource;
|
|
30
|
import eu.etaxonomy.cdm.model.description.DescriptionElementBase;
|
|
31
|
import eu.etaxonomy.cdm.model.description.Distribution;
|
|
32
|
import eu.etaxonomy.cdm.model.description.Feature;
|
|
33
|
import eu.etaxonomy.cdm.model.description.PresenceAbsenceTermBase;
|
|
34
|
import eu.etaxonomy.cdm.model.description.PresenceTerm;
|
|
35
|
import eu.etaxonomy.cdm.model.description.TaxonDescription;
|
|
36
|
import eu.etaxonomy.cdm.model.location.NamedArea;
|
|
37
|
import eu.etaxonomy.cdm.model.location.TdwgArea;
|
|
38
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
|
39
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
40
|
import eu.etaxonomy.cdm.model.taxon.Synonym;
|
|
41
|
import eu.etaxonomy.cdm.model.taxon.Taxon;
|
|
42
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
43
|
|
|
44
|
/**
|
|
45
|
* @author a.mueller
|
|
46
|
* @created 29.05.2008
|
|
47
|
* @version 1.0
|
|
48
|
*/
|
|
49
|
@Component
|
|
50
|
public class TcsRdfTaxonImport extends TcsRdfImportBase implements ICdmIO<TcsRdfImportState> {
|
|
51
|
private static final Logger logger = Logger.getLogger(TcsRdfTaxonImport.class);
|
|
52
|
|
|
53
|
private static int modCount = 30000;
|
|
54
|
|
|
55
|
public TcsRdfTaxonImport(){
|
|
56
|
super();
|
|
57
|
}
|
|
58
|
|
|
59
|
|
|
60
|
@Override
|
|
61
|
public boolean doCheck(TcsRdfImportState state){
|
|
62
|
boolean result = true;
|
|
63
|
logger.warn("Checking for Taxa not yet implemented");
|
|
64
|
//result &= checkArticlesWithoutJournal(bmiConfig);
|
|
65
|
//result &= checkPartOfJournal(bmiConfig);
|
|
66
|
|
|
67
|
return result;
|
|
68
|
}
|
|
69
|
|
|
70
|
protected static CdmSingleAttributeXmlMapperBase[] standardMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
71
|
// new CdmTextElementMapper("genusPart", "genusOrUninomial")
|
|
72
|
|
|
73
|
};
|
|
74
|
|
|
75
|
|
|
76
|
protected static CdmSingleAttributeXmlMapperBase[] operationalMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
77
|
new CdmUnclearMapper("hasName")
|
|
78
|
,new CdmUnclearMapper("hasName")
|
|
79
|
, new CdmUnclearMapper("accordingTo")
|
|
80
|
, new CdmUnclearMapper("hasRelationship")
|
|
81
|
, new CdmUnclearMapper("code", nsTgeo)
|
|
82
|
};
|
|
83
|
|
|
84
|
protected static CdmSingleAttributeXmlMapperBase[] unclearMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
85
|
new CdmUnclearMapper("primary")
|
|
86
|
, new CdmUnclearMapper("note", nsTcom)
|
|
87
|
, new CdmUnclearMapper("taxonStatus", nsTpalm)
|
|
88
|
|
|
89
|
, new CdmUnclearMapper("TaxonName", nsTn)
|
|
90
|
, new CdmUnclearMapper("dateOfEntry", nsTpalm)
|
|
91
|
};
|
|
92
|
|
|
93
|
|
|
94
|
|
|
95
|
@Override
|
|
96
|
protected void doInvoke(TcsRdfImportState state){
|
|
97
|
|
|
98
|
MapWrapper<TaxonBase> taxonMap = (MapWrapper<TaxonBase>)state.getStore(ICdmIO.TAXON_STORE);
|
|
99
|
MapWrapper<TaxonNameBase> taxonNameMap = (MapWrapper<TaxonNameBase>)state.getStore(ICdmIO.TAXONNAME_STORE);
|
|
100
|
MapWrapper<Reference> referenceMap = (MapWrapper<Reference>)state.getStore(ICdmIO.REFERENCE_STORE);
|
|
101
|
MapWrapper<Reference> nomRefMap = (MapWrapper<Reference>)state.getStore(ICdmIO.NOMREF_STORE);
|
|
102
|
|
|
103
|
String xmlElementName;
|
|
104
|
String xmlAttributeName;
|
|
105
|
Namespace elementNamespace;
|
|
106
|
Namespace attributeNamespace;
|
|
107
|
|
|
108
|
logger.info("start makeTaxa ...");
|
|
109
|
|
|
110
|
TcsRdfImportConfigurator config = state.getConfig();
|
|
111
|
Element root = config.getSourceRoot();
|
|
112
|
|
|
113
|
Namespace rdfNamespace = config.getRdfNamespace();
|
|
114
|
|
|
115
|
String idNamespace = "TaxonConcept";
|
|
116
|
xmlElementName = "TaxonConcept";
|
|
117
|
elementNamespace = config.getTcNamespace();
|
|
118
|
List<Element> elTaxonConcepts = root.getChildren(xmlElementName, elementNamespace);
|
|
119
|
|
|
120
|
ITaxonService taxonService = getTaxonService();
|
|
121
|
|
|
122
|
//debug names
|
|
123
|
if (false){
|
|
124
|
for (Object nameResource : taxonNameMap.keySet()){
|
|
125
|
System.out.println(nameResource);
|
|
126
|
}
|
|
127
|
}
|
|
128
|
|
|
129
|
int i = 0;
|
|
130
|
//for each taxonConcept
|
|
131
|
for (Element elTaxonConcept : elTaxonConcepts){
|
|
132
|
if ((i++ % modCount) == 0 && i > 1){ logger.info("Taxa handled: " + (i-1));}
|
|
133
|
|
|
134
|
//
|
|
135
|
String taxonAbout = elTaxonConcept.getAttributeValue("about", rdfNamespace);
|
|
136
|
|
|
137
|
//hasName
|
|
138
|
xmlElementName = "hasName";
|
|
139
|
elementNamespace = config.getTcNamespace();
|
|
140
|
xmlAttributeName = "resource";
|
|
141
|
attributeNamespace = rdfNamespace;
|
|
142
|
String strNameResource= XmlHelp.getChildAttributeValue(elTaxonConcept, xmlElementName, elementNamespace, xmlAttributeName, attributeNamespace);
|
|
143
|
TaxonNameBase taxonNameBase = taxonNameMap.get(strNameResource);
|
|
144
|
if (taxonNameBase == null){
|
|
145
|
logger.warn("Taxon has no name: " + taxonAbout + "; Resource: " + strNameResource);
|
|
146
|
}
|
|
147
|
|
|
148
|
|
|
149
|
//accordingTo
|
|
150
|
xmlElementName = "accordingTo";
|
|
151
|
elementNamespace = config.getTcNamespace();
|
|
152
|
xmlAttributeName = "resource";
|
|
153
|
attributeNamespace = rdfNamespace;
|
|
154
|
//String strAccordingTo = elTaxonConcept.getChildTextTrim(xmlElementName, elementNamespace);
|
|
155
|
String strAccordingTo = XmlHelp.getChildAttributeValue(elTaxonConcept, xmlElementName, elementNamespace, xmlAttributeName, attributeNamespace);
|
|
156
|
|
|
157
|
|
|
158
|
// //FIXME
|
|
159
|
// String secId = "pub_999999";
|
|
160
|
Reference sec = referenceMap.get(strAccordingTo);
|
|
161
|
if (sec == null){
|
|
162
|
sec = nomRefMap.get(strAccordingTo);
|
|
163
|
}
|
|
164
|
if (sec == null){
|
|
165
|
logger.warn("sec could not be found in referenceMap or nomRefMap for secId: " + strAccordingTo);
|
|
166
|
}
|
|
167
|
|
|
168
|
TaxonBase taxonBase;
|
|
169
|
Namespace geoNamespace = config.getGeoNamespace();
|
|
170
|
if (hasIsSynonymRelation(elTaxonConcept, rdfNamespace) || isSynonym(elTaxonConcept, config.getPalmNamespace())){
|
|
171
|
//Synonym
|
|
172
|
taxonBase = Synonym.NewInstance(taxonNameBase, sec);
|
|
173
|
List<DescriptionElementBase> geo = makeGeo(elTaxonConcept, geoNamespace, rdfNamespace);
|
|
174
|
if (geo.size() > 0){
|
|
175
|
logger.warn("Synonym (" + taxonAbout + ") has geo description!");
|
|
176
|
}
|
|
177
|
}else{
|
|
178
|
//Taxon
|
|
179
|
Taxon taxon = Taxon.NewInstance(taxonNameBase, sec);
|
|
180
|
List<DescriptionElementBase> geoList = makeGeo(elTaxonConcept, geoNamespace, rdfNamespace);
|
|
181
|
TaxonDescription description = TaxonDescription.NewInstance(taxon);
|
|
182
|
description.addDescriptionSource(taxon.getSec());
|
|
183
|
for (DescriptionElementBase geo: geoList){
|
|
184
|
description.addElement(geo);
|
|
185
|
DescriptionElementSource source = DescriptionElementSource.NewInstance(null, null, taxon.getSec(), null);
|
|
186
|
geo.addSource(source);
|
|
187
|
}
|
|
188
|
taxon.addDescription(description);
|
|
189
|
taxonBase = taxon;
|
|
190
|
}
|
|
191
|
|
|
192
|
Set<String> omitAttributes = null;
|
|
193
|
makeStandardMapper(elTaxonConcept, taxonBase, omitAttributes, standardMappers);
|
|
194
|
|
|
195
|
ImportHelper.setOriginalSource(taxonBase, config.getSourceReference(), taxonAbout, idNamespace);
|
|
196
|
checkAdditionalContents(elTaxonConcept, standardMappers, operationalMappers, unclearMappers);
|
|
197
|
|
|
198
|
taxonMap.put(taxonAbout, taxonBase);
|
|
199
|
|
|
200
|
}
|
|
201
|
//invokeRelations(source, cdmApp, deleteAll, taxonMap, referenceMap);
|
|
202
|
logger.info("saving " + taxonMap.size()+ " taxa ...");
|
|
203
|
taxonService.save(taxonMap.objects());
|
|
204
|
logger.info("end makeTaxa ...");
|
|
205
|
return;
|
|
206
|
}
|
|
207
|
|
|
208
|
|
|
209
|
/**
|
|
210
|
* @param rdfNamespace
|
|
211
|
* @param elTaxonConcept
|
|
212
|
* @return
|
|
213
|
*/
|
|
214
|
private boolean isSynonym(Element elTaxonConcept, Namespace tpalmNamespace) {
|
|
215
|
if (elTaxonConcept == null || ! "TaxonConcept".equalsIgnoreCase(elTaxonConcept.getName()) ){
|
|
216
|
return false;
|
|
217
|
}
|
|
218
|
Element status = elTaxonConcept.getChild("taxonStatus", tpalmNamespace);
|
|
219
|
if (status == null){
|
|
220
|
return false;
|
|
221
|
}else{
|
|
222
|
String statusText = status.getTextNormalize();
|
|
223
|
if ("S".equalsIgnoreCase(statusText)){
|
|
224
|
return true;
|
|
225
|
}else if ("A".equalsIgnoreCase(statusText)){
|
|
226
|
return false;
|
|
227
|
}else if ("C".equalsIgnoreCase(statusText)){
|
|
228
|
return false;
|
|
229
|
}else if ("V".equalsIgnoreCase(statusText)){
|
|
230
|
return false;
|
|
231
|
}else if ("O".equalsIgnoreCase(statusText)){
|
|
232
|
return false;
|
|
233
|
}else if ("U".equalsIgnoreCase(statusText)){
|
|
234
|
return false;
|
|
235
|
}else{
|
|
236
|
logger.warn("Unknown taxon status: " + statusText);
|
|
237
|
return false;
|
|
238
|
}
|
|
239
|
}
|
|
240
|
}
|
|
241
|
|
|
242
|
|
|
243
|
private boolean hasIsSynonymRelation(Element elTaxonConcept, Namespace rdfNamespace){
|
|
244
|
boolean result = false;
|
|
245
|
if (elTaxonConcept == null || ! "TaxonConcept".equalsIgnoreCase(elTaxonConcept.getName()) ){
|
|
246
|
return false;
|
|
247
|
}
|
|
248
|
|
|
249
|
String elName = "relationshipCategory";
|
|
250
|
Filter filter = new ElementFilter(elName, elTaxonConcept.getNamespace());
|
|
251
|
Iterator<Element> relationshipCategories = elTaxonConcept.getDescendants(filter);
|
|
252
|
while (relationshipCategories.hasNext()){
|
|
253
|
Element relationshipCategory = relationshipCategories.next();
|
|
254
|
Attribute resource = relationshipCategory.getAttribute("resource", rdfNamespace);
|
|
255
|
String isSynonymFor = "http://rs.tdwg.org/ontology/voc/TaxonConcept#IsSynonymFor";
|
|
256
|
if (resource != null && isSynonymFor.equalsIgnoreCase(resource.getValue()) ){
|
|
257
|
return true;
|
|
258
|
}
|
|
259
|
}
|
|
260
|
return result;
|
|
261
|
}
|
|
262
|
|
|
263
|
private List<DescriptionElementBase> makeGeo(Element elConcept, Namespace geoNamespace, Namespace rdfNamespace){
|
|
264
|
List<DescriptionElementBase> result = new ArrayList<DescriptionElementBase>();
|
|
265
|
String xmlElementName = "code";
|
|
266
|
List<Element> elGeos = elConcept.getChildren(xmlElementName, geoNamespace);
|
|
267
|
|
|
268
|
int i = 0;
|
|
269
|
//for each geoTag
|
|
270
|
for (Element elGeo : elGeos){
|
|
271
|
//if ((i++ % modCount) == 0){ logger.info("Geocodes handled: " + (i-1));}
|
|
272
|
|
|
273
|
String strGeoRegion = elGeo.getAttributeValue("resource", rdfNamespace);
|
|
274
|
strGeoRegion = strGeoRegion.replace("http://rs.tdwg.org/ontology/voc/GeographicRegion#", "");
|
|
275
|
NamedArea namedArea = TdwgArea.getAreaByTdwgAbbreviation(strGeoRegion);
|
|
276
|
PresenceAbsenceTermBase status = PresenceTerm.PRESENT();
|
|
277
|
DescriptionElementBase distribution = Distribution.NewInstance(namedArea, status);
|
|
278
|
distribution.setFeature(Feature.DISTRIBUTION());
|
|
279
|
//System.out.println(namedArea);
|
|
280
|
|
|
281
|
result.add(distribution);
|
|
282
|
}
|
|
283
|
return result;
|
|
284
|
}
|
|
285
|
|
|
286
|
/* (non-Javadoc)
|
|
287
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
|
288
|
*/
|
|
289
|
protected boolean isIgnore(TcsRdfImportState state){
|
|
290
|
return ! state.getConfig().isDoTaxa();
|
|
291
|
}
|
|
292
|
|
|
293
|
|
|
294
|
}
|