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 static eu.etaxonomy.cdm.io.common.ImportHelper.NO_OVERWRITE;
|
13
|
import static eu.etaxonomy.cdm.io.common.ImportHelper.OBLIGATORY;
|
14
|
import static eu.etaxonomy.cdm.io.common.ImportHelper.OVERWRITE;
|
15
|
|
16
|
import java.util.ArrayList;
|
17
|
import java.util.HashSet;
|
18
|
import java.util.List;
|
19
|
import java.util.Set;
|
20
|
|
21
|
import org.apache.log4j.Logger;
|
22
|
import org.jdom.Element;
|
23
|
import org.jdom.Namespace;
|
24
|
import org.springframework.stereotype.Component;
|
25
|
|
26
|
import com.hp.hpl.jena.rdf.model.Model;
|
27
|
import com.hp.hpl.jena.rdf.model.Resource;
|
28
|
import com.hp.hpl.jena.rdf.model.Statement;
|
29
|
|
30
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
31
|
import eu.etaxonomy.cdm.common.XmlHelp;
|
32
|
import eu.etaxonomy.cdm.io.common.mapping.berlinModel.CdmOneToManyMapper;
|
33
|
import eu.etaxonomy.cdm.io.common.ICdmIO;
|
34
|
import eu.etaxonomy.cdm.io.common.IImportConfigurator;
|
35
|
import eu.etaxonomy.cdm.io.common.ImportHelper;
|
36
|
import eu.etaxonomy.cdm.io.common.MapWrapper;
|
37
|
import eu.etaxonomy.cdm.io.common.mapping.CdmSingleAttributeMapperBase;
|
38
|
import eu.etaxonomy.cdm.io.common.mapping.IRdfMapper;
|
39
|
import eu.etaxonomy.cdm.io.common.mapping.IXmlMapper;
|
40
|
import eu.etaxonomy.cdm.model.agent.Team;
|
41
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
42
|
import eu.etaxonomy.cdm.model.common.Marker;
|
43
|
import eu.etaxonomy.cdm.model.common.MarkerType;
|
44
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
45
|
import eu.etaxonomy.cdm.model.reference.IArticle;
|
46
|
import eu.etaxonomy.cdm.model.reference.IBookSection;
|
47
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
48
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
49
|
import eu.etaxonomy.cdm.model.reference.ReferenceType;
|
50
|
import eu.etaxonomy.cdm.strategy.exceptions.UnknownCdmTypeException;
|
51
|
|
52
|
/**
|
53
|
* @author a.mueller
|
54
|
* @created 29.05.2008
|
55
|
* @version 1.0
|
56
|
*/
|
57
|
@Component
|
58
|
public class TcsRdfReferenceImport extends TcsRdfImportBase implements ICdmIO<TcsRdfImportState> {
|
59
|
private static final Logger logger = Logger.getLogger(TcsRdfReferenceImport.class);
|
60
|
|
61
|
private static int modCount = 1000;
|
62
|
|
63
|
public TcsRdfReferenceImport(){
|
64
|
super();
|
65
|
}
|
66
|
|
67
|
@Override
|
68
|
public boolean doCheck(TcsRdfImportState state){
|
69
|
boolean result = true;
|
70
|
result &= checkArticlesWithoutJournal(state.getConfig());
|
71
|
//result &= checkPartOfJournal(config);
|
72
|
|
73
|
return result;
|
74
|
}
|
75
|
|
76
|
private static boolean checkArticlesWithoutJournal(IImportConfigurator bmiConfig){
|
77
|
try {
|
78
|
boolean result = true;
|
79
|
//TODO
|
80
|
// result = firstRow = false;
|
81
|
// }
|
82
|
//
|
83
|
return result;
|
84
|
} catch (Exception e) {
|
85
|
e.printStackTrace();
|
86
|
return false;
|
87
|
}
|
88
|
}
|
89
|
|
90
|
|
91
|
protected static IRdfMapper[] standardMappers = new IRdfMapper[]{
|
92
|
//new CdmTextElementMapper("edition", "edition"),
|
93
|
new CdmTextElementMapper("volume", "volume"),
|
94
|
new CdmTextElementMapper("placePublished", "placePublished"),
|
95
|
new CdmTextElementMapper("publisher", "publisher"),
|
96
|
//new CdmTextElementMapper("isbn", "isbn"),
|
97
|
new CdmTextElementMapper("pages", "pages"),
|
98
|
//new CdmTextElementMapper("series", "series"),
|
99
|
//new CdmTextElementMapper("issn", "issn"),
|
100
|
//new CdmTextElementMapper("url", "uri")
|
101
|
};
|
102
|
|
103
|
protected static CdmSingleAttributeRDFMapperBase[] operationalMappers = new CdmSingleAttributeRDFMapperBase[]{
|
104
|
new CdmUnclearMapper("year")
|
105
|
, new CdmUnclearMapper("title")
|
106
|
, new CdmUnclearMapper("shortTitle")
|
107
|
, new CdmUnclearMapper("publicationType")
|
108
|
, new CdmUnclearMapper("parentPublication")
|
109
|
, new CdmUnclearMapper("authorship")
|
110
|
|
111
|
};
|
112
|
|
113
|
// protected static String[] createdAndNotesAttributes = new String[]{
|
114
|
// "created_When", "updated_When", "created_Who", "updated_Who", "notes"
|
115
|
// };
|
116
|
|
117
|
protected static CdmSingleAttributeXmlMapperBase[] unclearMappers = new CdmSingleAttributeXmlMapperBase[]{
|
118
|
|
119
|
};
|
120
|
|
121
|
|
122
|
|
123
|
private boolean makeStandardMapper(Statement resource, Reference ref, Set<String> omitAttributes){
|
124
|
if (omitAttributes == null){
|
125
|
omitAttributes = new HashSet<String>();
|
126
|
}
|
127
|
boolean result = true;
|
128
|
for (IRdfMapper mapper : standardMappers){
|
129
|
if (mapper instanceof CdmSingleAttributeMapperBase){
|
130
|
makeSingleAttributeMapper((CdmSingleAttributeRDFMapperBase)mapper, resource, ref, omitAttributes);
|
131
|
}else if (mapper instanceof CdmOneToManyMapper){
|
132
|
makeMultipleAttributeMapper((CdmOneToManyMapper)mapper, resource, ref, omitAttributes);
|
133
|
}else{
|
134
|
logger.error("Unrecognized mapper type");
|
135
|
return false;
|
136
|
}
|
137
|
|
138
|
|
139
|
}
|
140
|
return true;
|
141
|
}
|
142
|
|
143
|
private boolean makeSingleAttributeMapper(CdmSingleAttributeRDFMapperBase mapper, Statement resource, Reference ref, Set<String> omitAttributes){
|
144
|
boolean result = true;
|
145
|
Object value = getValue(mapper, resource);
|
146
|
//write to destination
|
147
|
if (value != null){
|
148
|
String destinationAttribute = mapper.getDestinationAttribute();
|
149
|
if (! omitAttributes.contains(destinationAttribute)){
|
150
|
result &= ImportHelper.addValue(value, ref, destinationAttribute, mapper.getTypeClass(), OVERWRITE, OBLIGATORY);
|
151
|
}
|
152
|
}
|
153
|
return result;
|
154
|
}
|
155
|
|
156
|
private boolean makeMultipleAttributeMapper(CdmOneToManyMapper<?,?,CdmTextElementMapper> mapper, Statement parentElement, Reference ref, Set<String> omitAttributes){
|
157
|
if (omitAttributes == null){
|
158
|
omitAttributes = new HashSet<String>();
|
159
|
}
|
160
|
boolean result = true;
|
161
|
String destinationAttribute = mapper.getSingleAttributeName();
|
162
|
List<Object> sourceValues = new ArrayList<Object>();
|
163
|
List<Class> classes = new ArrayList<Class>();
|
164
|
for (CdmTextElementMapper singleMapper : mapper.getSingleMappers()){
|
165
|
String sourceAttribute = singleMapper.getSourceAttribute();
|
166
|
Object value = getValue(singleMapper, parentElement);
|
167
|
//Object value = valueMap.get(sourceAttribute);
|
168
|
sourceValues.add(value);
|
169
|
Class clazz = singleMapper.getTypeClass();
|
170
|
classes.add(clazz);
|
171
|
}
|
172
|
|
173
|
result &= ImportHelper.addMultipleValues(sourceValues, ref, destinationAttribute, classes, NO_OVERWRITE, OBLIGATORY);
|
174
|
return result;
|
175
|
}
|
176
|
|
177
|
// private Object getValue(CdmSingleAttributeRDFMapperBase mapper, Element parentElement){
|
178
|
// String sourceAttribute = mapper.getSourceAttribute().toLowerCase();
|
179
|
// Namespace sourceNamespace = mapper.getSourceNamespace(parentElement);
|
180
|
// Element child = parentElement.getChild(sourceAttribute, sourceNamespace);
|
181
|
// if (child == null){
|
182
|
// return null;
|
183
|
// }
|
184
|
// if (child.getContentSize() > 1){
|
185
|
// logger.warn("Element is not String");
|
186
|
// }
|
187
|
// Object value = child.getTextTrim();
|
188
|
// return value;
|
189
|
// }
|
190
|
|
191
|
@Override
|
192
|
protected void doInvoke(TcsRdfImportState state){
|
193
|
MapWrapper<Reference> referenceMap = (MapWrapper<Reference>)state.getStore(ICdmIO.REFERENCE_STORE);
|
194
|
MapWrapper<Reference> nomRefMap = (MapWrapper<Reference>)state.getStore(ICdmIO.NOMREF_STORE);
|
195
|
|
196
|
TcsRdfImportConfigurator config = state.getConfig();
|
197
|
Model root = config.getSourceRoot();
|
198
|
logger.info("start makeReferences ...");
|
199
|
|
200
|
String tcsElementName;
|
201
|
Namespace tcsNamespace;
|
202
|
|
203
|
String rdfNamespace = config.getRdfNamespaceURIString();
|
204
|
String publicationNamespace = config.getPublicationNamespaceURI();
|
205
|
|
206
|
String idNamespace = "PublicationCitation";
|
207
|
tcsElementName = "PublicationCitation";
|
208
|
|
209
|
/*List<Element> elPublicationCitations = root.l
|
210
|
|
211
|
int nomRefCount = 0;
|
212
|
int biblioRefsCount = 0;
|
213
|
|
214
|
int i = 0;
|
215
|
//for each publication
|
216
|
for (Element elPublicationCitation : elPublicationCitations){
|
217
|
|
218
|
if ((++i % modCount) == 0){ logger.info("references handled: " + (i-1));}
|
219
|
|
220
|
//create publication element
|
221
|
String strAbout = elPublicationCitation.getAttributeValue("about", rdfNamespace);
|
222
|
|
223
|
tcsElementName = "publicationType";
|
224
|
tcsNamespace = publicationNamespace;
|
225
|
String strPubType = XmlHelp.getChildAttributeValue(elPublicationCitation, tcsElementName, tcsNamespace, "resource", rdfNamespace);
|
226
|
|
227
|
try {
|
228
|
Reference<?> ref = TcsRdfTransformer.pubTypeStr2PubType(strPubType);
|
229
|
if (ref==null){
|
230
|
ref = ReferenceFactory.newGeneric();
|
231
|
}
|
232
|
|
233
|
Set<String> omitAttributes = null;
|
234
|
makeStandardMapper(elPublicationCitation, ref, omitAttributes);
|
235
|
|
236
|
|
237
|
tcsElementName = "authorship";
|
238
|
tcsNamespace = publicationNamespace;
|
239
|
String strAuthorship = elPublicationCitation.getChildText(tcsElementName, tcsNamespace);
|
240
|
//TODO
|
241
|
TeamOrPersonBase<?> authorship = Team.NewInstance();
|
242
|
authorship.setTitleCache(strAuthorship, true);
|
243
|
ref.setAuthorship(authorship);
|
244
|
|
245
|
tcsElementName = "year";
|
246
|
tcsNamespace = publicationNamespace;
|
247
|
String strYear = elPublicationCitation.getChildText(tcsElementName, tcsNamespace);
|
248
|
TimePeriod datePublished = ImportHelper.getDatePublished(strYear);
|
249
|
ref.setDatePublished(datePublished);
|
250
|
|
251
|
//Reference
|
252
|
//TODO
|
253
|
tcsElementName = "parentPublication";
|
254
|
tcsNamespace = publicationNamespace;
|
255
|
String strParent = XmlHelp.getChildAttributeValue(elPublicationCitation, tcsElementName, tcsNamespace, "resource", rdfNamespace);
|
256
|
Reference<?> parent = referenceMap.get(strParent);
|
257
|
if (parent != null){
|
258
|
if ((ref.getType().equals(ReferenceType.Article)) && (parent.getType().equals(ReferenceType.Journal))){
|
259
|
((IArticle)ref).setInJournal(parent);
|
260
|
}else if ((ref.getType().equals(ReferenceType.BookSection)) && (parent.getType().equals(ReferenceType.Book))){
|
261
|
((IBookSection)ref).setInBook(parent);
|
262
|
}else{
|
263
|
logger.warn("parent type (parent: " + parent.getType() +", child("+strAbout+"): " + ref.getType() + ")not yet implemented");
|
264
|
//ref.setParent(parent);
|
265
|
}
|
266
|
}
|
267
|
|
268
|
|
269
|
//FIXME
|
270
|
//nomRef and reference
|
271
|
tcsElementName = "shortTitle";
|
272
|
tcsNamespace = publicationNamespace;
|
273
|
boolean nomRefExists = false;
|
274
|
String strShortTitle = elPublicationCitation.getChildText(tcsElementName, tcsNamespace);
|
275
|
if (! CdmUtils.Nz(strShortTitle).trim().equals("")){
|
276
|
ref.setTitle(strShortTitle);
|
277
|
ImportHelper.setOriginalSource(ref, config.getSourceReference(), strAbout, idNamespace);
|
278
|
nomRefMap.put(strAbout, ref);
|
279
|
nomRefCount++;
|
280
|
nomRefExists = true;
|
281
|
}
|
282
|
|
283
|
tcsElementName = "title";
|
284
|
tcsNamespace = publicationNamespace;
|
285
|
String strTitle = elPublicationCitation.getChildText(tcsElementName, tcsNamespace);
|
286
|
tcsNamespace = publicationNamespace;
|
287
|
if (! CdmUtils.Nz(strTitle).trim().equals("") || nomRefExists == false){
|
288
|
//TODO
|
289
|
Reference<?> biblioRef = (Reference<?>)ref.clone();
|
290
|
biblioRef.setTitle(strTitle);
|
291
|
ImportHelper.setOriginalSource(biblioRef, config.getSourceReference(), strAbout, idNamespace);
|
292
|
referenceMap.put(strAbout, biblioRef);
|
293
|
biblioRefsCount++;
|
294
|
}
|
295
|
|
296
|
|
297
|
checkAdditionalContents(elPublicationCitation, standardMappers, operationalMappers, unclearMappers);
|
298
|
|
299
|
if (state.getConfig().isPublishReferences()){
|
300
|
ref.addMarker(Marker.NewInstance(MarkerType.IN_BIBLIOGRAPHY(), false));
|
301
|
}
|
302
|
|
303
|
//ImportHelper.setOriginalSource(nameBase, tcsConfig.getSourceReference(), nameId);
|
304
|
|
305
|
} catch (UnknownCdmTypeException e) {
|
306
|
//FIXME
|
307
|
logger.warn("Name with id " + strAbout + " has unknown type " + strPubType + " and could not be saved.");
|
308
|
state.setUnsuccessfull();
|
309
|
}
|
310
|
}
|
311
|
|
312
|
//change conceptRef uuid
|
313
|
Reference<?> sec = referenceMap.get(config.getSourceSecId());
|
314
|
if (sec == null){
|
315
|
sec = nomRefMap.get(config.getSourceSecId());
|
316
|
}
|
317
|
if (sec != null){
|
318
|
sec.setUuid(config.getSecUuid());
|
319
|
logger.info("concept reference uuid changed to: " + config.getSecUuid());
|
320
|
}
|
321
|
|
322
|
//save and store in map
|
323
|
logger.info("Save nomenclatural references (" + nomRefCount + ")");
|
324
|
getReferenceService().save(nomRefMap.objects());
|
325
|
logger.info("Save bibliographical references (" + biblioRefsCount +")");
|
326
|
getReferenceService().save(referenceMap.objects());
|
327
|
|
328
|
//referenceService.saveReferenceAll(referenceMap.objects());
|
329
|
logger.info("end makeReferences ...");*/
|
330
|
return;
|
331
|
|
332
|
}
|
333
|
|
334
|
/* (non-Javadoc)
|
335
|
* @see eu.etaxonomy.cdm.io.common.CdmIoBase#isIgnore(eu.etaxonomy.cdm.io.common.IImportConfigurator)
|
336
|
*/
|
337
|
@Override
|
338
|
protected boolean isIgnore(TcsRdfImportState state){
|
339
|
return (state.getConfig().getDoReferences() == IImportConfigurator.DO_REFERENCES.NONE);
|
340
|
}
|
341
|
|
342
|
}
|