|
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.tcsrdf;
|
|
11
|
|
|
12
|
import java.util.List;
|
|
13
|
import java.util.Set;
|
|
14
|
|
|
15
|
import org.apache.log4j.Logger;
|
|
16
|
import org.jdom.Attribute;
|
|
17
|
import org.jdom.Element;
|
|
18
|
import org.jdom.Namespace;
|
|
19
|
import org.springframework.stereotype.Component;
|
|
20
|
|
|
21
|
import eu.etaxonomy.cdm.common.XmlHelp;
|
|
22
|
import eu.etaxonomy.cdm.io.common.ICdmIO;
|
|
23
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
|
24
|
import eu.etaxonomy.cdm.io.common.MapWrapper;
|
|
25
|
import eu.etaxonomy.cdm.model.agent.INomenclaturalAuthor;
|
|
26
|
import eu.etaxonomy.cdm.model.agent.Team;
|
|
27
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
|
28
|
import eu.etaxonomy.cdm.model.common.Marker;
|
|
29
|
import eu.etaxonomy.cdm.model.common.MarkerType;
|
|
30
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
|
31
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
|
32
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatus;
|
|
33
|
import eu.etaxonomy.cdm.model.name.NomenclaturalStatusType;
|
|
34
|
import eu.etaxonomy.cdm.model.name.NonViralName;
|
|
35
|
import eu.etaxonomy.cdm.model.name.Rank;
|
|
36
|
import eu.etaxonomy.cdm.model.name.TaxonNameBase;
|
|
37
|
import eu.etaxonomy.cdm.model.reference.IGeneric;
|
|
38
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
|
39
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
|
40
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
|
41
|
|
|
42
|
/**
|
|
43
|
* @author a.mueller
|
|
44
|
* @created 29.05.2008
|
|
45
|
* @version 1.0
|
|
46
|
*/
|
|
47
|
@Component
|
|
48
|
public class TcsRdfTaxonNameImport extends TcsRdfImportBase implements ICdmIO<TcsRdfImportState> {
|
|
49
|
private static final Logger logger = Logger.getLogger(TcsRdfTaxonNameImport.class);
|
|
50
|
|
|
51
|
private static int modCount = 5000;
|
|
52
|
|
|
53
|
public TcsRdfTaxonNameImport(){
|
|
54
|
super();
|
|
55
|
}
|
|
56
|
|
|
57
|
@Override
|
|
58
|
public boolean doCheck(TcsRdfImportState config){
|
|
59
|
boolean result = true;
|
|
60
|
logger.warn("BasionymRelations not yet implemented");
|
|
61
|
logger.warn("Checking for TaxonNames not yet implemented");
|
|
62
|
//result &= checkArticlesWithoutJournal(tcsConfig);
|
|
63
|
//result &= checkPartOfJournal(tcsConfig);
|
|
64
|
|
|
65
|
return result;
|
|
66
|
}
|
|
67
|
|
|
68
|
protected static CdmSingleAttributeXmlMapperBase[] standardMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
69
|
new CdmTextElementMapper("genusPart", "genusOrUninomial")
|
|
70
|
, new CdmTextElementMapper("uninomial", "genusOrUninomial") //TODO make it a more specific Mapper for both attributes
|
|
71
|
, new CdmTextElementMapper("specificEpithet", "specificEpithet")
|
|
72
|
, new CdmTextElementMapper("infraspecificEpithet", "infraSpecificEpithet")
|
|
73
|
, new CdmTextElementMapper("infragenericEpithet", "infraGenericEpithet")
|
|
74
|
, new CdmTextElementMapper("microReference", nsTcom, "nomenclaturalMicroReference")
|
|
75
|
|
|
76
|
};
|
|
77
|
|
|
78
|
protected static CdmSingleAttributeXmlMapperBase[] operationalMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
79
|
new CdmUnclearMapper("basionymAuthorship")
|
|
80
|
, new CdmUnclearMapper("combinationAuthorship")
|
|
81
|
, new CdmUnclearMapper("hasAnnotation")
|
|
82
|
, new CdmUnclearMapper("rank")
|
|
83
|
, new CdmUnclearMapper("nomenclaturalCode")
|
|
84
|
, new CdmUnclearMapper("publishedIn", nsTcom)
|
|
85
|
, new CdmUnclearMapper("year")
|
|
86
|
};
|
|
87
|
|
|
88
|
protected static CdmSingleAttributeXmlMapperBase[] unclearMappers = new CdmSingleAttributeXmlMapperBase[]{
|
|
89
|
new CdmUnclearMapper("authorship")
|
|
90
|
, new CdmUnclearMapper("rankString")
|
|
91
|
, new CdmUnclearMapper("nameComplete")
|
|
92
|
, new CdmUnclearMapper("hasBasionym")
|
|
93
|
, new CdmUnclearMapper("dateOfEntry", nsTpalm)
|
|
94
|
};
|
|
95
|
|
|
96
|
@Override
|
|
97
|
protected void doInvoke(TcsRdfImportState state){
|
|
98
|
|
|
99
|
MapWrapper<TaxonNameBase> taxonNameMap = (MapWrapper<TaxonNameBase>)state.getStore(ICdmIO.TAXONNAME_STORE);
|
|
100
|
MapWrapper<Reference> referenceMap = (MapWrapper<Reference>)state.getStore(ICdmIO.REFERENCE_STORE);
|
|
101
|
MapWrapper<TeamOrPersonBase> authorMap = (MapWrapper<TeamOrPersonBase>)state.getStore(ICdmIO.TEAM_STORE);
|
|
102
|
|
|
103
|
String tcsElementName;
|
|
104
|
Namespace tcsNamespace;
|
|
105
|
String value;
|
|
106
|
|
|
107
|
logger.info("start makeTaxonNames ...");
|
|
108
|
TcsRdfImportConfigurator config = state.getConfig();
|
|
109
|
Element root = config.getSourceRoot();
|
|
110
|
|
|
111
|
Namespace rdfNamespace = config.getRdfNamespace();
|
|
112
|
Namespace taxonNameNamespace = config.getTnNamespace();
|
|
113
|
|
|
114
|
String idNamespace = "TaxonName";
|
|
115
|
|
|
116
|
List<Element> elTaxonNames = root.getChildren("TaxonName", taxonNameNamespace);
|
|
117
|
|
|
118
|
int i = 0;
|
|
119
|
//for each taxonName
|
|
120
|
for (Element elTaxonName : elTaxonNames){
|
|
121
|
|
|
122
|
if ((++i % modCount) == 0){ logger.info("Names handled: " + (i-1));}
|
|
123
|
|
|
124
|
Attribute about = elTaxonName.getAttribute("about", rdfNamespace);
|
|
125
|
|
|
126
|
//create TaxonName element
|
|
127
|
String nameAbout = elTaxonName.getAttributeValue("about", rdfNamespace);
|
|
128
|
String strRank = XmlHelp.getChildAttributeValue(elTaxonName, "rank", taxonNameNamespace, "resource", rdfNamespace);
|
|
129
|
String strNomenclaturalCode = XmlHelp.getChildAttributeValue(elTaxonName, "nomenclaturalCode", taxonNameNamespace, "resource", rdfNamespace);
|
|
130
|
|
|
131
|
try {
|
|
132
|
Rank rank = TcsRdfTransformer.rankString2Rank(strRank);
|
|
133
|
NomenclaturalCode nomCode = TcsRdfTransformer.nomCodeString2NomCode(strNomenclaturalCode);
|
|
134
|
TaxonNameBase nameBase = nomCode.getNewTaxonNameInstance(rank);
|
|
135
|
|
|
136
|
Set<String> omitAttributes = null;
|
|
137
|
makeStandardMapper(elTaxonName, nameBase, omitAttributes, standardMappers);
|
|
138
|
|
|
139
|
//Reference
|
|
140
|
//TODO
|
|
141
|
tcsElementName = "publishedIn";
|
|
142
|
tcsNamespace = config.getCommonNamespace();
|
|
143
|
value = (String)ImportHelper.getXmlInputValue(elTaxonName, tcsElementName, tcsNamespace);
|
|
144
|
if (value != null){
|
|
145
|
ReferenceFactory refFactory = ReferenceFactory.newInstance();
|
|
146
|
IGeneric nomRef = refFactory.newGeneric(); //TODO
|
|
147
|
nomRef.setTitleCache(value, true);
|
|
148
|
nameBase.setNomenclaturalReference(nomRef);
|
|
149
|
|
|
150
|
//TODO
|
|
151
|
tcsElementName = "year";
|
|
152
|
tcsNamespace = taxonNameNamespace;
|
|
153
|
Integer year = null;
|
|
154
|
value = (String)ImportHelper.getXmlInputValue(elTaxonName, tcsElementName, tcsNamespace);
|
|
155
|
if (value != null){
|
|
156
|
try {
|
|
157
|
year = Integer.valueOf(value);
|
|
158
|
TimePeriod timeP = TimePeriod.NewInstance(year);
|
|
159
|
nomRef.setDatePublished(timeP);
|
|
160
|
} catch (RuntimeException e) {
|
|
161
|
logger.warn("year could not be parsed");
|
|
162
|
}
|
|
163
|
}
|
|
164
|
if (state.getConfig().isPublishReferences()){
|
|
165
|
((Reference)nomRef).addMarker(Marker.NewInstance(MarkerType.PUBLISH(), false));
|
|
166
|
}
|
|
167
|
}
|
|
168
|
|
|
169
|
//Status
|
|
170
|
tcsNamespace = taxonNameNamespace;
|
|
171
|
Element elAnnotation = elTaxonName.getChild("hasAnnotation", tcsNamespace);
|
|
172
|
if (elAnnotation != null){
|
|
173
|
Element elNomenclaturalNote = elAnnotation.getChild("NomenclaturalNote", tcsNamespace);
|
|
174
|
if (elNomenclaturalNote != null){
|
|
175
|
String statusValue = (String)ImportHelper.getXmlInputValue(elNomenclaturalNote, "note", tcsNamespace);
|
|
176
|
String type = XmlHelp.getChildAttributeValue(elNomenclaturalNote, "type", tcsNamespace, "resource", rdfNamespace);
|
|
177
|
String tdwgType = "http://rs.tdwg.org/ontology/voc/TaxonName#PublicationStatus";
|
|
178
|
if (tdwgType.equalsIgnoreCase(type)){
|
|
179
|
try {
|
|
180
|
NomenclaturalStatusType statusType = TcsRdfTransformer.nomStatusString2NomStatus(statusValue);
|
|
181
|
//NomenclaturalStatusType statusType = NomenclaturalStatusType.getNomenclaturalStatusTypeByAbbreviation(statusValue);
|
|
182
|
if (statusType != null){
|
|
183
|
nameBase.addStatus(NomenclaturalStatus.NewInstance(statusType));
|
|
184
|
}
|
|
185
|
} catch (UnknownCdmTypeException e) {
|
|
186
|
if (! statusValue.equals("valid")){
|
|
187
|
logger.warn("Unknown NomenclaturalStatusType: " + statusValue);
|
|
188
|
}
|
|
189
|
}
|
|
190
|
}
|
|
191
|
}
|
|
192
|
}
|
|
193
|
|
|
194
|
if (nameBase instanceof NonViralName){
|
|
195
|
NonViralName<?> nonViralName = (NonViralName<?>)nameBase;
|
|
196
|
|
|
197
|
//AuthorTeams
|
|
198
|
//TODO
|
|
199
|
tcsElementName = "basionymAuthorship";
|
|
200
|
String basionymAuthorValue = (String)ImportHelper.getXmlInputValue(elTaxonName, tcsElementName, taxonNameNamespace);
|
|
201
|
if (basionymAuthorValue != null){
|
|
202
|
INomenclaturalAuthor basionymAuthor = Team.NewInstance();
|
|
203
|
basionymAuthor.setNomenclaturalTitle(basionymAuthorValue);
|
|
204
|
nonViralName.setBasionymAuthorTeam(basionymAuthor);
|
|
205
|
}
|
|
206
|
|
|
207
|
//TODO
|
|
208
|
tcsElementName = "combinationAuthorship";
|
|
209
|
String combinationAuthorValue = (String)ImportHelper.getXmlInputValue(elTaxonName, tcsElementName, taxonNameNamespace);
|
|
210
|
if (combinationAuthorValue != null){
|
|
211
|
INomenclaturalAuthor combinationAuthor = Team.NewInstance();
|
|
212
|
combinationAuthor.setNomenclaturalTitle(combinationAuthorValue);
|
|
213
|
nonViralName.setCombinationAuthorTeam(combinationAuthor);
|
|
214
|
}
|
|
215
|
|
|
216
|
//set the authorshipCache
|
|
217
|
tcsElementName = "authorship";
|
|
218
|
String authorValue = (String)ImportHelper.getXmlInputValue(elTaxonName, tcsElementName, taxonNameNamespace);
|
|
219
|
String cache = nonViralName.getAuthorshipCache();
|
|
220
|
if ( authorValue != null){
|
|
221
|
//compare existing authorship cache with new one and check if it is necessary to
|
|
222
|
//make cache protected //TODO refinement
|
|
223
|
if (cache == null){
|
|
224
|
nonViralName.setAuthorshipCache(authorValue);
|
|
225
|
}else{
|
|
226
|
cache = basionymAuthorValue == null ? cache : cache.replace(basionymAuthorValue, "");
|
|
227
|
cache = combinationAuthorValue == null ? cache : cache.replace(combinationAuthorValue, "");
|
|
228
|
cache = cache.replace("\\(|\\)", "");
|
|
229
|
cache = cache.trim();
|
|
230
|
if (! cache.equals("")){
|
|
231
|
nonViralName.setAuthorshipCache(authorValue);
|
|
232
|
}
|
|
233
|
}
|
|
234
|
}
|
|
235
|
}
|
|
236
|
ImportHelper.setOriginalSource(nameBase, config.getSourceReference(), nameAbout, idNamespace);
|
|
237
|
|
|
238
|
checkAdditionalContents(elTaxonName, standardMappers, operationalMappers, unclearMappers);
|
|
239
|
|
|
240
|
//nameId
|
|
241
|
//TODO
|
|
242
|
//ImportHelper.setOriginalSource(nameBase, tcsConfig.getSourceReference(), nameId);
|
|
243
|
taxonNameMap.put(nameAbout, nameBase);
|
|
244
|
|
|
245
|
} catch (UnknownCdmTypeException e) {
|
|
246
|
//FIXME
|
|
247
|
logger.warn("Name with id " + nameAbout + " has unknown rank " + strRank + " and could not be saved.");
|
|
248
|
state.setUnsuccessfull();
|
|
249
|
}
|
|
250
|
}
|
|
251
|
logger.info(i + " names handled");
|
|
252
|
getNameService().save(taxonNameMap.objects());
|
|
253
|
// makeNameSpecificData(nameMap);
|
|
254
|
logger.info("end makeTaxonNames ...");
|
|
255
|
return;
|
|
256
|
|
|
257
|
}
|
|
258
|
/* (non-Javadoc)
|
|
259
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
|
260
|
*/
|
|
261
|
protected boolean isIgnore(TcsRdfImportState state){
|
|
262
|
return ! state.getConfig().isDoTaxonNames();
|
|
263
|
}
|
|
264
|
|
|
265
|
}
|