1
|
package eu.dnetlib.parthenos.rdf;
|
2
|
|
3
|
import java.util.Iterator;
|
4
|
import java.util.List;
|
5
|
import java.util.Set;
|
6
|
|
7
|
import com.google.common.collect.Iterators;
|
8
|
import com.google.common.collect.Lists;
|
9
|
import eu.dnetlib.parthenos.CRM;
|
10
|
import eu.dnetlib.parthenos.CRMpe;
|
11
|
import eu.dnetlib.parthenos.catalogue.CatalogueLicense;
|
12
|
import org.apache.commons.logging.Log;
|
13
|
import org.apache.commons.logging.LogFactory;
|
14
|
import org.apache.jena.assembler.AssemblerHelp;
|
15
|
import org.apache.jena.rdf.model.*;
|
16
|
import org.apache.jena.vocabulary.RDF;
|
17
|
import org.apache.jena.vocabulary.RDFS;
|
18
|
import org.springframework.stereotype.Component;
|
19
|
|
20
|
/**
|
21
|
* Created by Alessia Bardi on 11/12/2017.
|
22
|
*
|
23
|
* @author Alessia Bardi
|
24
|
*/
|
25
|
@Component
|
26
|
public class ResourceReader {
|
27
|
|
28
|
private static final Log log = LogFactory.getLog(ResourceReader.class);
|
29
|
//sparql template for services include <availability> field that is not explicit in the model
|
30
|
private Property availibilityProperty = ResourceFactory.createProperty("","availability");
|
31
|
//sparql template for services include <activitytype> field that is not explicit in the model
|
32
|
private Property activitytypeProperty = ResourceFactory.createProperty("","activitytype");
|
33
|
//sparql template for datasets include <encoding> field that is generated from the creation event
|
34
|
private Property encodingProperty = ResourceFactory.createProperty("","encoding");
|
35
|
//sparql template for datasets include <used_software> field that is generated from the creation event
|
36
|
private Property usedSoftwareProperty = ResourceFactory.createProperty("","used_software");
|
37
|
//sparql template for software include <usedby> field that is generated from the creation event of the resource using the software
|
38
|
private Property usedByProperty = ResourceFactory.createProperty("","usedby");
|
39
|
//sparql template for datasets include <creator> field that is generated from the creation event
|
40
|
private Property creatorProperty = ResourceFactory.createProperty("","creator");
|
41
|
|
42
|
|
43
|
public String getTitle(final Resource resource) {
|
44
|
final Statement s = resource.getProperty(CRM.P102_has_title);
|
45
|
if (s != null) {
|
46
|
RDFNode obj = s.getObject();
|
47
|
if(obj.isLiteral()) return obj.asLiteral().getLexicalForm();
|
48
|
}
|
49
|
//if we do not find the crm:P102_has_title, let's get the label
|
50
|
return getLabel(resource);
|
51
|
}
|
52
|
|
53
|
public Iterator<String> getDescriptions(final Resource resource) {
|
54
|
StmtIterator it = resource.listProperties(CRM.P3_has_note);
|
55
|
return Iterators.transform(it, f -> f.getString().replace("'", "\'"));
|
56
|
}
|
57
|
|
58
|
public String getLabel(final Resource resource) {
|
59
|
if(resource == null) return "";
|
60
|
if (resource.hasProperty(RDFS.label)) {
|
61
|
return resource.getProperty(RDFS.label).getString().replace("'", "\'");
|
62
|
} else return "";
|
63
|
}
|
64
|
|
65
|
//NOTE: based on the sparql templates competency is a string, in the model it is a PE36_Competency_Type
|
66
|
public Iterator<String> getCompetences(final Resource resource) {
|
67
|
StmtIterator it = resource.listProperties(CRMpe.PP45_has_competency);
|
68
|
return Iterators.transform(it, f -> f.getString());
|
69
|
}
|
70
|
|
71
|
public String getAvailability(final Resource resource) {
|
72
|
String availability = "";
|
73
|
if (resource.hasProperty(availibilityProperty)){
|
74
|
availability = resource.getProperty(availibilityProperty).getString();
|
75
|
}
|
76
|
return availability;
|
77
|
}
|
78
|
|
79
|
public String getConditionOfUse(final Resource resource) {
|
80
|
StmtIterator it = resource.listProperties(CRM.P16_used_specific_object);
|
81
|
while(it.hasNext()){
|
82
|
Resource obj = it.next().getResource();
|
83
|
if(obj.hasProperty(RDF.type, CRM.E30_Right) && obj.hasProperty(CRM.P3_has_note)){
|
84
|
String rightsString = obj.getProperty(CRM.P3_has_note).getString();
|
85
|
if(obj.hasProperty(CRM.P2_has_type)){
|
86
|
Resource rightType = obj.getPropertyResourceValue(CRM.P2_has_type);
|
87
|
rightsString += " ["+getLabel(rightType)+"]";
|
88
|
}
|
89
|
return rightsString;
|
90
|
}
|
91
|
}
|
92
|
return "";
|
93
|
}
|
94
|
|
95
|
public CatalogueLicense getCatalogueLicense(final Resource resource){
|
96
|
if(resource.hasProperty(CRM.P16_used_specific_object)) {
|
97
|
Resource obj = resource.getPropertyResourceValue(CRM.P16_used_specific_object);
|
98
|
if (obj.hasProperty(CRM.P2_has_type)) {
|
99
|
String license = getLabel(obj.getPropertyResourceValue(CRM.P2_has_type));
|
100
|
return CatalogueLicense.getCatalogueLicenseFor(license);
|
101
|
}
|
102
|
}
|
103
|
return CatalogueLicense.NotSpecified;
|
104
|
}
|
105
|
|
106
|
public Iterator<String> getRDFClassNames(final Resource resource){
|
107
|
StmtIterator it = resource.listProperties(RDF.type);
|
108
|
return Iterators.transform(it, f -> f.getResource().getLocalName());
|
109
|
}
|
110
|
|
111
|
public Iterator<String> getActivityTypes(final Resource resource){
|
112
|
StmtIterator it = resource.listProperties(activitytypeProperty);
|
113
|
return Iterators.transform(it, f -> f.getString());
|
114
|
}
|
115
|
|
116
|
public Iterator<String> getProviderNames(final Resource resource){
|
117
|
StmtIterator sit = resource.listProperties(CRMpe.PP2_provided_by);
|
118
|
StmtIterator sit2 = resource.listProperties(CRMpe.PP25_has_maintaining_RI);
|
119
|
return Iterators.transform(Iterators.concat(sit, sit2), f -> {
|
120
|
Resource provider = f.getResource();
|
121
|
return getTitle(provider);
|
122
|
});
|
123
|
}
|
124
|
|
125
|
public Iterator<String> getProviderUris(final Resource resource){
|
126
|
StmtIterator sit = resource.listProperties(CRMpe.PP2_provided_by);
|
127
|
StmtIterator sit2 = resource.listProperties(CRMpe.PP25_has_maintaining_RI);
|
128
|
return Iterators.transform(Iterators.concat(sit, sit2), f -> {
|
129
|
return f.getResource().getURI();
|
130
|
});
|
131
|
}
|
132
|
|
133
|
public Iterator<String> getProviderContactPoints(final Resource resource){
|
134
|
StmtIterator it = resource.listProperties(CRMpe.PP2_provided_by);
|
135
|
return Iterators.transform(it, f -> {
|
136
|
Resource provider = f.getResource();
|
137
|
if (provider.hasProperty(CRM.P76_has_contact_point)) {
|
138
|
Resource contactPoint = provider.getPropertyResourceValue(CRM.P76_has_contact_point);
|
139
|
return getLabel(contactPoint);
|
140
|
}
|
141
|
else return "";
|
142
|
});
|
143
|
}
|
144
|
|
145
|
public Iterator<String> getResourceDirectContactPointsURI(final Resource resource){
|
146
|
StmtIterator it = resource.listProperties(CRM.P76_has_contact_point);
|
147
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
148
|
}
|
149
|
|
150
|
|
151
|
|
152
|
public Iterator<String> getHostedStuff(final Resource resource){
|
153
|
//In inference we trust
|
154
|
StmtIterator sit4 = resource.listProperties(CRMpe.PP4_hosts_object);
|
155
|
return Iterators.transform(sit4, f -> f.getResource().getURI());
|
156
|
// StmtIterator sit6 = resource.listProperties(CRMpe.PP6_hosts_digital_object);
|
157
|
// StmtIterator sit7 = resource.listProperties(CRMpe.PP7_hosts_software_object);
|
158
|
// StmtIterator sit8 = resource.listProperties(CRMpe.PP8_hosts_dataset);
|
159
|
// Iterator<String> it4 = Iterators.transform(sit4, f -> f.getResource().getURI());
|
160
|
// Iterator<String> it6 = Iterators.transform(sit6, f -> f.getResource().getURI());
|
161
|
// Iterator<String> it7 = Iterators.transform(sit7, f -> f.getResource().getURI());
|
162
|
// Iterator<String> it8 = Iterators.transform(sit8, f -> f.getResource().getURI());
|
163
|
// return Iterators.concat(it4,it6, it7, it8);
|
164
|
}
|
165
|
|
166
|
public Iterator<String> getHostedBys(final Resource resource){
|
167
|
//In inference we trust
|
168
|
StmtIterator sit4 = resource.listProperties(CRMpe.PP4i_is_object_hosted_by);
|
169
|
return Iterators.transform(sit4, f -> f.getResource().getURI());
|
170
|
// StmtIterator sit6 = resource.listProperties(CRMpe.PP6i_is_digital_object_hosted_by);
|
171
|
// StmtIterator sit7 = resource.listProperties(CRMpe.PP7i_is_software_object_hosted_by);
|
172
|
// StmtIterator sit8 = resource.listProperties(CRMpe.PP8i_is_dataset_hosted_by);
|
173
|
// Iterator<String> it4 = Iterators.transform(sit4, f -> f.getResource().getURI());
|
174
|
// Iterator<String> it6 = Iterators.transform(sit6, f -> f.getResource().getURI());
|
175
|
// Iterator<String> it7 = Iterators.transform(sit7, f -> f.getResource().getURI());
|
176
|
// Iterator<String> it8 = Iterators.transform(sit8, f -> f.getResource().getURI());
|
177
|
// return Iterators.concat(it4, it6, it7, it8);
|
178
|
}
|
179
|
|
180
|
|
181
|
public Iterator<String> getCuratedObjects(final Resource resource){
|
182
|
//In inference we trust
|
183
|
StmtIterator sit32 = resource.listProperties(CRMpe.PP32_curates);
|
184
|
return Iterators.transform(sit32, f -> f.getResource().getURI());
|
185
|
// StmtIterator sit11 = resource.listProperties(CRMpe.PP11_curates_volatile_digital_object);
|
186
|
// StmtIterator sit12 = resource.listProperties(CRMpe.PP12_curates_volatile_software);
|
187
|
// StmtIterator sit13 = resource.listProperties(CRMpe.PP13_curates_volatile_dataset);
|
188
|
// Iterator<String> it32 = Iterators.transform(sit32, f -> f.getResource().getURI());
|
189
|
// Iterator<String> it11 = Iterators.transform(sit11, f -> f.getResource().getURI());
|
190
|
// Iterator<String> it12 = Iterators.transform(sit12, f -> f.getResource().getURI());
|
191
|
// Iterator<String> it13 = Iterators.transform(sit13, f -> f.getResource().getURI());
|
192
|
// return Iterators.concat(it32, it11, it12, it13);
|
193
|
}
|
194
|
|
195
|
public Iterator<String> getCuratorUrls(final Resource resource){
|
196
|
//In inference we trust
|
197
|
StmtIterator sit32 = resource.listProperties(CRMpe.PP32i_is_curated_by);
|
198
|
return Iterators.transform(sit32, f -> f.getResource().getURI());
|
199
|
// StmtIterator sit11 = resource.listProperties(CRMpe.PP11i_is_volatile_digital_object_curated_by);
|
200
|
// StmtIterator sit12 = resource.listProperties(CRMpe.PP12i_is_volatile_software_curated_by);
|
201
|
// StmtIterator sit13 = resource.listProperties(CRMpe.PP13i_is_volatile_dataset_curated_by);
|
202
|
// Iterator<String> it32 = Iterators.transform(sit32, f -> f.getResource().getURI());
|
203
|
// Iterator<String> it11 = Iterators.transform(sit11, f -> f.getResource().getURI());
|
204
|
// Iterator<String> it12 = Iterators.transform(sit12, f -> f.getResource().getURI());
|
205
|
// Iterator<String> it13 = Iterators.transform(sit13, f -> f.getResource().getURI());
|
206
|
// return Iterators.concat(it32, it11, it12, it13);
|
207
|
}
|
208
|
|
209
|
public Iterator<String> getDeliversOnRequest(final Resource resource){
|
210
|
StmtIterator sit = resource.listProperties(CRMpe.PP15_delivers_on_request);
|
211
|
return Iterators.transform(sit, f -> f.getResource().getURI());
|
212
|
}
|
213
|
|
214
|
public Iterator<String> getRunsOnRequest(final Resource resource){
|
215
|
StmtIterator sit = resource.listProperties(CRMpe.PP14_runs_on_request);
|
216
|
return Iterators.transform(sit, f -> f.getResource().getURI());
|
217
|
}
|
218
|
|
219
|
public Iterator<String> getAccessPoints(final Resource resource){
|
220
|
StmtIterator it = resource.listProperties(CRMpe.PP28_has_designated_access_point);
|
221
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
222
|
}
|
223
|
|
224
|
public Iterator<String> getDeclarativeTimes(final Resource resource){
|
225
|
StmtIterator it = resource.listProperties(CRMpe.PP42_has_declarative_time);
|
226
|
return Iterators.transform(it, f -> f.getString());
|
227
|
}
|
228
|
|
229
|
public Iterator<String> getProtocols(final Resource resource){
|
230
|
StmtIterator it = resource.listProperties(CRMpe.PP29_uses_access_protocol);
|
231
|
return Iterators.transform(it, f -> getLabel(f.getResource()));
|
232
|
}
|
233
|
|
234
|
public Iterator<String> getCurationPlans(final Resource resource){
|
235
|
StmtIterator it = resource.listProperties(CRMpe.PP31_uses_curation_plan);
|
236
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
237
|
}
|
238
|
|
239
|
public Iterator<String> getMemberUrls(final Resource resource){
|
240
|
StmtIterator it = resource.listProperties(CRM.P107_has_current_or_former_member);
|
241
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
242
|
}
|
243
|
|
244
|
public Iterator<String> isMemberOf(final Resource resource){
|
245
|
StmtIterator it = resource.listProperties(CRM.P107i_is_current_or_former_member_of);
|
246
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
247
|
}
|
248
|
|
249
|
public Iterator<String> getProvidedServiceUrls(final Resource resource){
|
250
|
StmtIterator it = resource.listProperties(CRMpe.PP2i_provides);
|
251
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
252
|
}
|
253
|
|
254
|
public Iterator<String> getHasTypeLabels(final Resource resource){
|
255
|
StmtIterator it = resource.listProperties(CRM.P2_has_type);
|
256
|
return Iterators.transform(it, f -> getLabel(f.getResource()));
|
257
|
}
|
258
|
|
259
|
public Iterator<String> getIsPartOfUrls(final Resource resource){
|
260
|
//in inference we trust
|
261
|
//StmtIterator it = resource.listProperties(CRMpe.PP23i_is_dataset_part_of);
|
262
|
StmtIterator it = resource.listProperties(CRM.P106i_forms_part_of);
|
263
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
264
|
}
|
265
|
|
266
|
public Iterator<String> getHasPartUrls(final Resource resource){
|
267
|
//in inference we trust
|
268
|
//StmtIterator it = resource.listProperties(CRMpe.PP23_has_dataset_part);
|
269
|
StmtIterator it = resource.listProperties(CRM.P106_is_composed_of);
|
270
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
271
|
}
|
272
|
|
273
|
public Iterator<String> getSubjects(final Resource resource){
|
274
|
StmtIterator it = resource.listProperties(CRM.P129_is_about);
|
275
|
return Iterators.transform(it, f -> getLabel(f.getResource()));
|
276
|
}
|
277
|
|
278
|
public List<String> getTemporalCoverages(final Resource resource){
|
279
|
return getCoverages(resource, CRM.E4_Period);
|
280
|
}
|
281
|
|
282
|
public List<String> getSpatialCoverages(final Resource resource){
|
283
|
return getCoverages(resource, CRM.E53_Place);
|
284
|
}
|
285
|
|
286
|
private List<String> getCoverages(final Resource resource, final Resource coverageType){
|
287
|
List<String> cov = Lists.newArrayList();
|
288
|
StmtIterator it = resource.listProperties(CRM.P129_is_about);
|
289
|
while(it.hasNext()){
|
290
|
Resource r = it.next().getResource();
|
291
|
if(r.hasProperty(RDF.type, coverageType)){
|
292
|
cov.add(getLabel(r));
|
293
|
}
|
294
|
}
|
295
|
return cov;
|
296
|
}
|
297
|
|
298
|
public Iterator<String> getEncodings(final Resource resource){
|
299
|
StmtIterator it = resource.listProperties(encodingProperty);
|
300
|
return Iterators.transform(it, f -> f.getString());
|
301
|
}
|
302
|
|
303
|
public Iterator<String> getCreatorsURIs(final Resource resource){
|
304
|
StmtIterator it = resource.listProperties(creatorProperty);
|
305
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
306
|
}
|
307
|
|
308
|
public Iterator<String> getMetadata(final Resource resource){
|
309
|
StmtIterator it = resource.listProperties(CRMpe.PP39i_has_metadata);
|
310
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
311
|
}
|
312
|
public Iterator<String> getDescribedDataset(final Resource resource){
|
313
|
StmtIterator it = resource.listProperties(CRMpe.PP39_is_metadata_for);
|
314
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
315
|
}
|
316
|
|
317
|
public Iterator<String> getOfferedServiceUrls(final Resource resource){
|
318
|
StmtIterator it = resource.listProperties(CRMpe.PP1_currently_offers);
|
319
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
320
|
}
|
321
|
|
322
|
public String getStartTime(final Resource resource){
|
323
|
Resource timespan = resource.getPropertyResourceValue(CRM.P4_has_time_span);
|
324
|
if(timespan != null){
|
325
|
if(timespan.hasProperty(CRM.P82a_begin_of_the_begin))
|
326
|
return timespan.getProperty(CRM.P82a_begin_of_the_begin).getString();
|
327
|
}
|
328
|
return "";
|
329
|
}
|
330
|
|
331
|
public Iterator<String> getMaintainerUrls(final Resource resource){
|
332
|
StmtIterator it = resource.listProperties(CRMpe.PP44_has_maintaining_team);
|
333
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
334
|
}
|
335
|
|
336
|
public List<String> getMaintainerContacts(final Resource resource){
|
337
|
List<String> res = Lists.newArrayList();
|
338
|
StmtIterator it = resource.listProperties(CRMpe.PP44_has_maintaining_team);
|
339
|
while(it.hasNext()){
|
340
|
Resource maintainer = it.next().getResource();
|
341
|
Iterator<String> itM = getResourceDirectContactPointsURI(maintainer);
|
342
|
while(itM.hasNext()){
|
343
|
res.add(itM.next());
|
344
|
}
|
345
|
}
|
346
|
return res;
|
347
|
}
|
348
|
|
349
|
public Iterator<String> getMaintainersForManagementSection(final Resource resource){
|
350
|
StmtIterator it = resource.listProperties(CRMpe.PP44_has_maintaining_team);
|
351
|
return Iterators.transform(it, f -> getLabel(f.getResource()));
|
352
|
}
|
353
|
|
354
|
public Iterator<String> getMaintainedUrls(final Resource resource){
|
355
|
StmtIterator it = resource.listProperties(CRMpe.PP44i_is_maintaining_team_of);
|
356
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
357
|
}
|
358
|
|
359
|
public Iterator<String> getSnapshots(final Resource resource){
|
360
|
StmtIterator it = resource.listProperties(CRMpe.PP17_has_snapshot);
|
361
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
362
|
}
|
363
|
|
364
|
public Iterator<String> getIsSnapshotOfs(final Resource resource){
|
365
|
StmtIterator it = resource.listProperties(CRMpe.PP17i_is_snapshot_of);
|
366
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
367
|
}
|
368
|
|
369
|
public Iterator<String> getUsedSoftware(final Resource resource){
|
370
|
StmtIterator it = resource.listProperties(usedSoftwareProperty);
|
371
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
372
|
}
|
373
|
|
374
|
public Iterator<String> getUsedBy(final Resource resource){
|
375
|
StmtIterator it = resource.listProperties(usedByProperty);
|
376
|
return Iterators.transform(it, f -> f.getResource().getURI());
|
377
|
}
|
378
|
|
379
|
//NOTE: languages are resources in CRM but our sparql template make them strings by taking the label.
|
380
|
public Iterator<String> getLanguages(final Resource resource){
|
381
|
StmtIterator it = resource.listProperties(CRM.P72_has_language);
|
382
|
return Iterators.transform(it, f -> f.getString());
|
383
|
}
|
384
|
|
385
|
/**
|
386
|
* Finds the most specific type of res.
|
387
|
*
|
388
|
* @param res Resource you want to find the most specific type
|
389
|
* @param fallbackType Resource representing the type to return if there is no type or if we get AmbiguousSpecificTypeException
|
390
|
* @return Resource: the most specific type, if any. fallbackType otherwise
|
391
|
*/
|
392
|
public Resource findSpecificType(final Resource res, final Resource fallbackType) {
|
393
|
Resource type = fallbackType;
|
394
|
Set<Resource> types = AssemblerHelp.findSpecificTypes(res, fallbackType);
|
395
|
if (types == null || types.isEmpty()) {
|
396
|
log.warn("No specific type found. Returning the fallback type: " + fallbackType);
|
397
|
}
|
398
|
if (types.size() == 1) {
|
399
|
type = types.iterator().next();
|
400
|
}
|
401
|
if (types.size() > 1) {
|
402
|
log.warn("Found more than one possible specific type: choosing the first");
|
403
|
types.stream().forEach((t) -> log.warn(t));
|
404
|
type = types.iterator().next();
|
405
|
}
|
406
|
return type;
|
407
|
}
|
408
|
}
|