1
|
package eu.dnetlib.parthenos.catalogue;
|
2
|
|
3
|
import java.io.BufferedOutputStream;
|
4
|
import java.io.ByteArrayOutputStream;
|
5
|
import java.io.IOException;
|
6
|
import java.net.URISyntaxException;
|
7
|
import java.util.Iterator;
|
8
|
import java.util.List;
|
9
|
import java.util.Set;
|
10
|
|
11
|
import com.fasterxml.jackson.core.JsonEncoding;
|
12
|
import com.fasterxml.jackson.core.JsonFactory;
|
13
|
import com.fasterxml.jackson.core.JsonGenerator;
|
14
|
import com.google.common.base.Joiner;
|
15
|
import com.google.common.collect.Iterators;
|
16
|
import com.google.common.collect.Lists;
|
17
|
import com.google.common.collect.Sets;
|
18
|
import eu.dnetlib.parthenos.CRM;
|
19
|
import eu.dnetlib.parthenos.publisher.ParthenosPublisherException;
|
20
|
import eu.dnetlib.parthenos.rdf.ResourceReader;
|
21
|
import org.apache.commons.lang3.StringUtils;
|
22
|
import org.apache.commons.logging.Log;
|
23
|
import org.apache.commons.logging.LogFactory;
|
24
|
import org.apache.jena.rdf.model.RDFNode;
|
25
|
import org.apache.jena.rdf.model.Resource;
|
26
|
import org.apache.jena.rdf.model.StmtIterator;
|
27
|
import org.apache.jena.vocabulary.RDF;
|
28
|
import org.springframework.beans.factory.annotation.Autowired;
|
29
|
import org.springframework.stereotype.Component;
|
30
|
|
31
|
/**
|
32
|
* Created by Alessia Bardi on 21/11/2017.
|
33
|
*
|
34
|
* @author Alessia Bardi
|
35
|
*/
|
36
|
@Component
|
37
|
public class CatalogueRegistrator {
|
38
|
|
39
|
private static final Log log = LogFactory.getLog(CatalogueRegistrator.class);
|
40
|
private final String PARTHENOS_BASE_URL = "http://parthenos.d4science.org";
|
41
|
|
42
|
@Autowired
|
43
|
private ResourceReader resourceReader;
|
44
|
|
45
|
@Autowired
|
46
|
private CatalogueAPIClient catalogueAPIClient;
|
47
|
|
48
|
public String register(final Resource resource, final Resource type, final String datasourceName)
|
49
|
throws IOException, ParthenosPublisherException, URISyntaxException, InterruptedException {
|
50
|
String resURI = resource.getURI();
|
51
|
log.debug(String.format("Catalogue --> Processing resource : %s with type: %s from source: %s", resURI, type.getLocalName(), datasourceName));
|
52
|
String resCatName = catalogueAPIClient.getNameForCatalogue(resURI.substring(resURI.lastIndexOf("handle/") + 7));
|
53
|
if(catalogueAPIClient.isRegistered(resCatName)){
|
54
|
log.debug(resCatName+ " is already registered");
|
55
|
String json = getJson(type, resource, resCatName, datasourceName);
|
56
|
catalogueAPIClient.doUpdate(json, resCatName);
|
57
|
}
|
58
|
else {
|
59
|
//resource not yet registered
|
60
|
String json = getJson(type, resource, resCatName, datasourceName);
|
61
|
if(!catalogueAPIClient.doRegister(json, resCatName)){
|
62
|
log.warn(String.format("%s could not be registered even the second time, giving up", resURI));
|
63
|
return null;
|
64
|
}
|
65
|
}
|
66
|
log.debug(String.format("%s registered on the catalogue with name: %s", resURI, resCatName));
|
67
|
return resCatName;
|
68
|
}
|
69
|
|
70
|
protected boolean purge(final String resCatName) throws URISyntaxException, ParthenosPublisherException {
|
71
|
return catalogueAPIClient.purgeItem(resCatName);
|
72
|
}
|
73
|
|
74
|
public int purgeAll(final int bulkSize) throws ParthenosPublisherException {
|
75
|
return catalogueAPIClient.purgeAll(bulkSize);
|
76
|
}
|
77
|
|
78
|
|
79
|
|
80
|
protected String getJson(final Resource type, final Resource resource, final String resNameForCatalogue, final String datasourceName)
|
81
|
throws IOException, ParthenosPublisherException {
|
82
|
switch (type.getLocalName()) {
|
83
|
case "E29_Design_or_Procedure":
|
84
|
return getJsonForDesignProcedure(resource, resNameForCatalogue, datasourceName);
|
85
|
case "D14_Software":
|
86
|
return getJsonForSoftware(resource, resNameForCatalogue, datasourceName);
|
87
|
case "PE35_Project":
|
88
|
return getJsonForProject(resource, resNameForCatalogue, datasourceName);
|
89
|
case "PE1_Service":
|
90
|
return getJsonForService(resource, resNameForCatalogue, datasourceName);
|
91
|
case "E39_Actor":
|
92
|
return getJsonForActor(resource, resNameForCatalogue, datasourceName);
|
93
|
case "PE18_Dataset":
|
94
|
return getJsonForDataset(resource, resNameForCatalogue, datasourceName);
|
95
|
case "E78_Collection":
|
96
|
return getJsonForCollection(resource, resNameForCatalogue, datasourceName);
|
97
|
default:
|
98
|
throw new IllegalArgumentException(String.format("Type " + type.getLocalName() + " not supported"));
|
99
|
}
|
100
|
}
|
101
|
|
102
|
protected String getJsonForProject(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
103
|
throws IOException, ParthenosPublisherException {
|
104
|
JsonFactory jsonFactory = new JsonFactory();
|
105
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
106
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
107
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
108
|
jg.writeStartObject();
|
109
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
110
|
//Only RI_Project are linked to RIs that must be used as groups
|
111
|
List<String> maintainer_RIs = Lists.newArrayList(resourceReader.getMaintainersLabels(res));
|
112
|
if(maintainer_RIs.size() > 0) {
|
113
|
jg.writeArrayFieldStart("groups");
|
114
|
for(String ri : maintainer_RIs) {
|
115
|
String group = CKANUtils.getCkanGroup(ri);
|
116
|
if(StringUtils.isNotBlank(group)) {
|
117
|
jg.writeStartObject();
|
118
|
jg.writeStringField("name", group);
|
119
|
jg.writeEndObject();
|
120
|
}
|
121
|
}
|
122
|
jg.writeEndArray();
|
123
|
}
|
124
|
|
125
|
jg.writeStringField("maintainer", Joiner.on(", ").join(resourceReader.getMaintainersLabels(res)));
|
126
|
//TODO: it should be better to identify email contacts rather than generic contact labels of maintainer
|
127
|
//jg.writeStringField("maintainer_email", Joiner.on(", ").join(resourceReader.getMaintainerContacts(res)));
|
128
|
|
129
|
jg.writeArrayFieldStart("extras");
|
130
|
addExtra(jg, "system:type", CKANUtils.Project_type);
|
131
|
//specific class
|
132
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E7_Activity).getLocalName());
|
133
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL))
|
134
|
addExtra(jg, "Parthenos URL", res.getURI());
|
135
|
else addExtra(jg, "URL", res.getURI());
|
136
|
addIdentifiers(res, jg);
|
137
|
|
138
|
addExtra(jg, "started on", resourceReader.getStartTime(res));
|
139
|
int idx = 1;
|
140
|
Iterator<String> maintainers = resourceReader.getMaintainerUrls(res);
|
141
|
while(maintainers.hasNext()){
|
142
|
addExtra(jg, String.format("maintaining team (%d)", idx), maintainers.next());
|
143
|
idx++;
|
144
|
}
|
145
|
//addExtra(jg, "maintaining team", Joiner.on(", ").join(resourceReader.getMaintainerUrls(res)));
|
146
|
idx = 1;
|
147
|
Iterator<String> services = resourceReader.getOfferedServiceUrls(res);
|
148
|
while(services.hasNext()){
|
149
|
addExtra(jg, String.format("offers (%d)", idx), services.next());
|
150
|
idx++;
|
151
|
}
|
152
|
//addExtra(jg, "offers", Joiner.on(", ").join(resourceReader.getOfferedServiceUrls(res)));
|
153
|
|
154
|
jg.writeEndArray(); //end extras
|
155
|
|
156
|
jg.writeEndObject();
|
157
|
jg.close();
|
158
|
return out.toString("UTF-8");
|
159
|
}
|
160
|
|
161
|
protected String getJsonForService(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
162
|
throws IOException, ParthenosPublisherException {
|
163
|
JsonFactory jsonFactory = new JsonFactory();
|
164
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
165
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
166
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
167
|
jg.writeStartObject();
|
168
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
169
|
jg.writeStringField("maintainer", Joiner.on(", ").join(resourceReader.getMaintainersLabels(res)));
|
170
|
//TODO: it should be better to identify email contacts rather than generic contact labels of maintainer
|
171
|
//jg.writeStringField("maintainer_email", Joiner.on(", ").join(resourceReader.getMaintainerContacts(res)));
|
172
|
|
173
|
jg.writeArrayFieldStart("extras");
|
174
|
addExtra(jg, "system:type", CKANUtils.Service_type);
|
175
|
//specific class
|
176
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E7_Activity).getLocalName());
|
177
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL))
|
178
|
addExtra(jg, "Parthenos URL", res.getURI());
|
179
|
else addExtra(jg, "URL", res.getURI());
|
180
|
addIdentifiers(res, jg);
|
181
|
addExtra(jg, "competence", Joiner.on(", ").join(resourceReader.getCompetences(res)));
|
182
|
addExtra(jg, "activity type", Joiner.on(", ").join(resourceReader.getActivityTypes(res)));
|
183
|
//condition of use (Rights)
|
184
|
addExtra(jg, "condition of use", resourceReader.getConditionOfUse(res));
|
185
|
int idx = 1;
|
186
|
Iterator<String> contacts = Iterators.concat(resourceReader.getResourceDirectContactPointsURI(res), resourceReader.getProviderContactPoints(res));
|
187
|
while(contacts.hasNext()){
|
188
|
addExtra(jg, String.format("contact points (%d)", idx), contacts.next());
|
189
|
idx++;
|
190
|
}
|
191
|
//addExtra(jg, "contact points", Joiner.on(", ").join(Iterators.concat(resourceReader.getResourceDirectContactPointsURI(res), resourceReader.getProviderContactPoints(res))));
|
192
|
idx = 1;
|
193
|
Iterator<String> providers = resourceReader.getProviderUris(res);
|
194
|
while(providers.hasNext()){
|
195
|
addExtra(jg, String.format("provided by (%d)", idx), providers.next());
|
196
|
idx++;
|
197
|
}
|
198
|
//addExtra(jg, "provided by", Joiner.on(", ").join(resourceReader.getProviderUris(res)));
|
199
|
idx = 1;
|
200
|
Iterator<String> points = resourceReader.getAccessPoints(res);
|
201
|
while(points.hasNext()){
|
202
|
addExtra(jg, String.format("online access point (%d)", idx), points.next());
|
203
|
idx++;
|
204
|
}
|
205
|
//addExtra(jg, "online access point", Joiner.on(", ").join(resourceReader.getAccessPoints(res)));
|
206
|
addExtra(jg, "protocol", Joiner.on(", ").join(resourceReader.getProtocols(res)));
|
207
|
idx = 1;
|
208
|
Iterator<String> delivers = resourceReader.getDeliversOnRequest(res);
|
209
|
while(delivers.hasNext()){
|
210
|
addExtra(jg, String.format("delivers on request (%d)", idx), delivers.next());
|
211
|
idx++;
|
212
|
}
|
213
|
//addExtra(jg, "delivers on request", Joiner.on(", ").join(resourceReader.getDeliversOnRequest(res)));
|
214
|
idx = 1;
|
215
|
Iterator<String> runs = resourceReader.getRunsOnRequest(res);
|
216
|
while(runs.hasNext()){
|
217
|
addExtra(jg, String.format("runs on request (%d)", idx), runs.next());
|
218
|
idx++;
|
219
|
}
|
220
|
//addExtra(jg, "runs on request", Joiner.on(", ").join(resourceReader.getRunsOnRequest(res)));
|
221
|
idx = 1;
|
222
|
Iterator<String> hosts = resourceReader.getHostedStuff(res);
|
223
|
while(hosts.hasNext()){
|
224
|
addExtra(jg, String.format("hosts (%d)", idx), hosts.next());
|
225
|
idx++;
|
226
|
}
|
227
|
//addExtra(jg, "hosts", Joiner.on(", ").join(resourceReader.getHostedStuff(res)));
|
228
|
idx = 1;
|
229
|
Iterator<String> curates = resourceReader.getCuratedObjects(res);
|
230
|
while(curates.hasNext()){
|
231
|
addExtra(jg, String.format("curates (%d)", idx), curates.next());
|
232
|
idx++;
|
233
|
}
|
234
|
//addExtra(jg, "curates", Joiner.on(", ").join(resourceReader.getCuratedObjects(res)));
|
235
|
addExtra(jg, "declared begin/end of operation", Joiner.on(", ").join(resourceReader.getDeclarativeTimes(res)));
|
236
|
addExtra(jg, "availability", resourceReader.getAvailability(res));
|
237
|
idx = 1;
|
238
|
Iterator<String> plans = resourceReader.getCurationPlans(res);
|
239
|
while(plans.hasNext()){
|
240
|
addExtra(jg, String.format("uses curation plan (%d)", idx), plans.next());
|
241
|
idx++;
|
242
|
}
|
243
|
//addExtra(jg, "uses curation plan", Joiner.on(", ").join(resourceReader.getCurationPlans(res)));
|
244
|
addExtra(jg, "time of service", Joiner.on(", ").join(resourceReader.getDeclarativeTimes(res)));
|
245
|
//TODO: where to get it?
|
246
|
//addExtra(jg, "last confirmation", "");
|
247
|
//TODO: where to get it?
|
248
|
//addExtra(jg, "date of registration", "");
|
249
|
|
250
|
jg.writeEndArray(); //end extras
|
251
|
|
252
|
jg.writeEndObject();
|
253
|
jg.close();
|
254
|
return out.toString("UTF-8");
|
255
|
}
|
256
|
|
257
|
protected String getJsonForActor(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
258
|
throws IOException, ParthenosPublisherException {
|
259
|
JsonFactory jsonFactory = new JsonFactory();
|
260
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
261
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
262
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
263
|
jg.writeStartObject();
|
264
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
265
|
|
266
|
jg.writeArrayFieldStart("extras");
|
267
|
addExtra(jg, "system:type", CKANUtils.Actor_type);
|
268
|
//specific class
|
269
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E39_Actor).getLocalName());
|
270
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL))
|
271
|
addExtra(jg, "Parthenos URL", res.getURI());
|
272
|
else addExtra(jg, "URL", res.getURI());
|
273
|
addIdentifiers(res, jg);
|
274
|
addExtra(jg, "has type", Joiner.on(", ").join(resourceReader.getHasTypeLabels(res)));
|
275
|
int idx = 1;
|
276
|
Iterator<String> hasMembers = resourceReader.getMemberUrls(res);
|
277
|
while(hasMembers.hasNext()){
|
278
|
addExtra(jg, String.format("has member (%d)", idx), hasMembers.next());
|
279
|
idx++;
|
280
|
}
|
281
|
//addExtra(jg, "has member", Joiner.on(", ").join(resourceReader.getMemberUrls(res)));
|
282
|
idx = 1;
|
283
|
Iterator<String> isMembers = resourceReader.isMemberOf(res);
|
284
|
while(isMembers.hasNext()){
|
285
|
addExtra(jg, String.format("is member of (%d)", idx), isMembers.next());
|
286
|
idx++;
|
287
|
}
|
288
|
//addExtra(jg, "is member of", Joiner.on(", ").join(resourceReader.isMemberOf(res)));
|
289
|
idx = 1;
|
290
|
Iterator<String> provides = resourceReader.getProvidedServiceUrls(res);
|
291
|
while(provides.hasNext()){
|
292
|
addExtra(jg, String.format("provides (%d)", idx), provides.next());
|
293
|
idx++;
|
294
|
}
|
295
|
//addExtra(jg, "provides", Joiner.on(", ").join(resourceReader.getProvidedServiceUrls(res)));
|
296
|
|
297
|
idx = 1;
|
298
|
String contactPoints = "";
|
299
|
StmtIterator it = res.listProperties(CRM.P76_has_contact_point);
|
300
|
while(it.hasNext()) {
|
301
|
Resource cp = it.next().getResource();
|
302
|
Resource cpType = cp.getPropertyResourceValue(CRM.P2_has_type);
|
303
|
String cpTypeLabel = resourceReader.getLabel(cpType);
|
304
|
String cpLabel = resourceReader.getLabel(cp);
|
305
|
if (StringUtils.isNotBlank(cpLabel)) {
|
306
|
if (StringUtils.isNotBlank(cpTypeLabel)) {
|
307
|
addExtra(jg,String.format("contact point (%d) - %s ", idx, cpTypeLabel), cpLabel );
|
308
|
idx++;
|
309
|
//contactPoints += cpTypeLabel + ": ";
|
310
|
}
|
311
|
else{
|
312
|
addExtra(jg,String.format("contact point (%d)", idx), cpLabel );
|
313
|
idx++;
|
314
|
//contactPoints += cpLabel + "; ";
|
315
|
}
|
316
|
|
317
|
}
|
318
|
else{
|
319
|
addExtra(jg,String.format("contact point (%d)", idx), cp.getURI());
|
320
|
idx++;
|
321
|
}
|
322
|
}
|
323
|
// addExtra(jg,"contact points", contactPoints );
|
324
|
idx = 1;
|
325
|
Iterator<String> maintains = resourceReader.getMaintainedUrls(res);
|
326
|
while(maintains.hasNext()){
|
327
|
addExtra(jg, String.format("maintains (%d)", idx), maintains.next());
|
328
|
idx++;
|
329
|
}
|
330
|
//addExtra(jg, "maintains", Joiner.on(", ").join(resourceReader.getMaintainedUrls(res)));
|
331
|
|
332
|
jg.writeEndArray();
|
333
|
|
334
|
jg.writeEndObject();
|
335
|
jg.close();
|
336
|
return out.toString("UTF-8");
|
337
|
}
|
338
|
|
339
|
protected String getJsonForDataset(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
340
|
throws IOException, ParthenosPublisherException {
|
341
|
JsonFactory jsonFactory = new JsonFactory();
|
342
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
343
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
344
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
345
|
jg.writeStartObject();
|
346
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
347
|
|
348
|
jg.writeArrayFieldStart("extras");
|
349
|
addExtra(jg, "system:type", CKANUtils.Dataset_type);
|
350
|
//specific class
|
351
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E70_Thing).getLocalName());
|
352
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {
|
353
|
addExtra(jg, "Parthenos URL", res.getURI());
|
354
|
}
|
355
|
else addExtra(jg, "URL", res.getURI());
|
356
|
addIdentifiers(res, jg);
|
357
|
addExtra(jg, "has type", Joiner.on(", ").join(resourceReader.getHasTypeLabels(res)));
|
358
|
int idx = 1;
|
359
|
Iterator<String> isPartOf = resourceReader.getIsPartOfUrls(res);
|
360
|
while(isPartOf.hasNext()){
|
361
|
addExtra(jg, String.format("is part of (%d)", idx), isPartOf.next());
|
362
|
idx++;
|
363
|
}
|
364
|
//addExtra(jg, "is part of", Joiner.on(", ").join(resourceReader.getIsPartOfUrls(res)));
|
365
|
idx = 1;
|
366
|
Iterator<String> hasPart = resourceReader.getHasPartUrls(res);
|
367
|
while(hasPart.hasNext()){
|
368
|
addExtra(jg, String.format("has part (%d)", idx), hasPart.next());
|
369
|
idx++;
|
370
|
}
|
371
|
//addExtra(jg, "has part", Joiner.on(", ").join(resourceReader.getHasPartUrls(res)));
|
372
|
idx = 1;
|
373
|
Iterator<String> curators = resourceReader.getCuratorUrls(res);
|
374
|
while(curators.hasNext()){
|
375
|
addExtra(jg, String.format("curated by (%d)", idx), curators.next());
|
376
|
idx++;
|
377
|
}
|
378
|
//addExtra(jg, "curated by", Joiner.on(", ").join(resourceReader.getCuratorUrls(res)));
|
379
|
idx = 1;
|
380
|
Iterator<String> curationplans = resourceReader.getResourceCuratorCurationPlans(res);
|
381
|
while(curationplans.hasNext()){
|
382
|
addExtra(jg, String.format("curation plan (%d)", idx), curationplans.next());
|
383
|
idx++;
|
384
|
}
|
385
|
//addExtra(jg, "curation plan", Joiner.on(", ").join(resourceReader.getResourceCuratorCurationPlans(res)));
|
386
|
idx = 1;
|
387
|
Iterator<String> hostedbys = resourceReader.getHostedBys(res);
|
388
|
while(hostedbys.hasNext()){
|
389
|
addExtra(jg, String.format("hosted by (%d)", idx), hostedbys.next());
|
390
|
idx++;
|
391
|
}
|
392
|
//addExtra(jg, "hosted by", Joiner.on(", ").join(resourceReader.getHostedBys(res)));
|
393
|
addExtra(jg, "encoding type", Joiner.on(", ").join(resourceReader.getEncodings(res)));
|
394
|
idx = 1;
|
395
|
Iterator<String> creators = resourceReader.getCreatorsURIs(res);
|
396
|
while(creators.hasNext()){
|
397
|
addExtra(jg, String.format("creator (%d)", idx), creators.next());
|
398
|
idx++;
|
399
|
}
|
400
|
//addExtra(jg, "creator", Joiner.on(", ").join(resourceReader.getCreatorsURIs(res)));
|
401
|
idx = 1;
|
402
|
Iterator<String> sw = resourceReader.getUsedSoftware(res);
|
403
|
while(sw.hasNext()){
|
404
|
addExtra(jg, String.format("used software (%d)", idx), sw.next());
|
405
|
idx++;
|
406
|
}
|
407
|
|
408
|
addExtra(jg, "subject", Joiner.on(", ").join(resourceReader.getSubjects(res)));
|
409
|
addExtra(jg, "temporal coverage", Joiner.on(", ").join(resourceReader.getTemporalCoverages(res)));
|
410
|
addExtra(jg, "spatial coverage", Joiner.on(", ").join(resourceReader.getSpatialCoverages(res)));
|
411
|
idx = 1;
|
412
|
Iterator<String> usedbys = resourceReader.getUsedBy(res);
|
413
|
while(usedbys.hasNext()){
|
414
|
addExtra(jg, String.format("used by (%d)", idx), usedbys.next());
|
415
|
idx++;
|
416
|
}
|
417
|
//addExtra(jg, "used by", Joiner.on(", ").join(resourceReader.getUsedBy(res)));
|
418
|
addExtra(jg, "languages", Joiner.on(", ").join(resourceReader.getLanguages(res)));
|
419
|
idx = 1;
|
420
|
Iterator<String> metadata = resourceReader.getMetadata(res);
|
421
|
while(metadata.hasNext()){
|
422
|
addExtra(jg, String.format("has metadata (%d)", idx), metadata.next());
|
423
|
idx++;
|
424
|
}
|
425
|
//addExtra(jg, "has metadata", Joiner.on(", ").join(resourceReader.getMetadata(res)));
|
426
|
idx = 1;
|
427
|
Iterator<String> metadataFor = resourceReader.getDescribedDataset(res);
|
428
|
while(metadataFor.hasNext()){
|
429
|
addExtra(jg, String.format("is metadata for (%d)", idx), metadataFor.next());
|
430
|
idx++;
|
431
|
}
|
432
|
//addExtra(jg, "is metadata for", Joiner.on(", ").join(resourceReader.getDescribedDataset(res)));
|
433
|
idx = 1;
|
434
|
Iterator<String> snaphsots = resourceReader.getSnapshots(res);
|
435
|
while(snaphsots.hasNext()){
|
436
|
addExtra(jg, String.format("has snapshot (%d)", idx), snaphsots.next());
|
437
|
idx++;
|
438
|
}
|
439
|
//addExtra(jg, "has snapshot", Joiner.on(", ").join(resourceReader.getSnapshots(res)));
|
440
|
idx = 1;
|
441
|
Iterator<String> issnaphsots = resourceReader.getIsSnapshotOfs(res);
|
442
|
while(issnaphsots.hasNext()){
|
443
|
addExtra(jg, String.format("is snapshot of (%d)", idx), issnaphsots.next());
|
444
|
idx++;
|
445
|
}
|
446
|
//addExtra(jg, "is snapshot of", Joiner.on(", ").join(resourceReader.getIsSnapshotOfs(res)));
|
447
|
|
448
|
jg.writeEndArray();
|
449
|
|
450
|
jg.writeEndObject();
|
451
|
jg.close();
|
452
|
return out.toString("UTF-8");
|
453
|
}
|
454
|
|
455
|
protected String getJsonForSoftware(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
456
|
throws IOException, ParthenosPublisherException {
|
457
|
JsonFactory jsonFactory = new JsonFactory();
|
458
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
459
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
460
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
461
|
jg.writeStartObject();
|
462
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
463
|
|
464
|
jg.writeArrayFieldStart("extras");
|
465
|
addExtra(jg, "system:type", CKANUtils.Software_type);
|
466
|
//specific class
|
467
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E70_Thing).getLocalName());
|
468
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {
|
469
|
addExtra(jg, "Parthenos URL", res.getURI());
|
470
|
}
|
471
|
else addExtra(jg, "URL", res.getURI());
|
472
|
addIdentifiers(res, jg);
|
473
|
addExtra(jg, "subject", Joiner.on(", ").join(resourceReader.getSubjects(res)));
|
474
|
addExtra(jg, "has type", Joiner.on(", ").join(resourceReader.getHasTypeLabels(res)));
|
475
|
int idx = 1;
|
476
|
Iterator<String> hosted = resourceReader.getHostedBys(res);
|
477
|
while(hosted.hasNext()){
|
478
|
addExtra(jg, String.format("hosted by (%d)", idx), hosted.next());
|
479
|
idx++;
|
480
|
}
|
481
|
//addExtra(jg, "hosted by", Joiner.on(", ").join(resourceReader.getHostedBys(res)));
|
482
|
idx = 1;
|
483
|
Iterator<String> curated = resourceReader.getCuratorUrls(res);
|
484
|
while(curated.hasNext()){
|
485
|
addExtra(jg, String.format("curated by (%d)", idx), curated.next());
|
486
|
idx++;
|
487
|
}
|
488
|
//addExtra(jg, "curated by", Joiner.on(", ").join(resourceReader.getCuratorUrls(res)));
|
489
|
idx = 1;
|
490
|
Iterator<String> hasSnapshot = resourceReader.getSnapshots(res);
|
491
|
while(hasSnapshot.hasNext()){
|
492
|
addExtra(jg, String.format("has snapshot (%d)", idx), hasSnapshot.next());
|
493
|
idx++;
|
494
|
}
|
495
|
//addExtra(jg, "has snapshot", Joiner.on(", ").join(resourceReader.getSnapshots(res)));
|
496
|
idx = 1;
|
497
|
Iterator<String> isSnapshot = resourceReader.getIsSnapshotOfs(res);
|
498
|
while(isSnapshot.hasNext()){
|
499
|
addExtra(jg, String.format("is snapshot (%d)", idx), isSnapshot.next());
|
500
|
idx++;
|
501
|
}
|
502
|
//addExtra(jg, "is snapshot of", Joiner.on(", ").join(resourceReader.getIsSnapshotOfs(res)));
|
503
|
idx = 1;
|
504
|
Iterator<String> isPart = resourceReader.getIsPartOfUrls(res);
|
505
|
while(isPart.hasNext()){
|
506
|
addExtra(jg, String.format("is part of (%d)", idx), isPart.next());
|
507
|
idx++;
|
508
|
}
|
509
|
//addExtra(jg, "is part of", Joiner.on(", ").join(resourceReader.getIsPartOfUrls(res)));
|
510
|
idx = 1;
|
511
|
Iterator<String> hasPart = resourceReader.getHasPartUrls(res);
|
512
|
while(hasPart.hasNext()){
|
513
|
addExtra(jg, String.format("has part (%d)", idx), hasPart.next());
|
514
|
idx++;
|
515
|
}
|
516
|
//addExtra(jg, "has part", Joiner.on(", ").join(resourceReader.getHasPartUrls(res)));
|
517
|
idx = 1;
|
518
|
Iterator<String> hasRelease = resourceReader.getHasReleases(res);
|
519
|
while(hasRelease.hasNext()){
|
520
|
addExtra(jg, String.format("has release (%d)", idx), hasRelease.next());
|
521
|
idx++;
|
522
|
}
|
523
|
//addExtra(jg, "has release", Joiner.on(", ").join(resourceReader.getHasReleases(res)));
|
524
|
idx = 1;
|
525
|
Iterator<String> isRelease = resourceReader.getIsReleaseOfs(res);
|
526
|
while(isRelease.hasNext()){
|
527
|
addExtra(jg, String.format("is release (%d)", idx), isRelease.next());
|
528
|
idx++;
|
529
|
}
|
530
|
//addExtra(jg, "is release of", Joiner.on(", ").join(resourceReader.getIsReleaseOfs(res)));
|
531
|
idx = 1;
|
532
|
Iterator<String> used = resourceReader.getUsedBy(res);
|
533
|
while(used.hasNext()){
|
534
|
addExtra(jg, String.format("used by (%d)", idx), used.next());
|
535
|
idx++;
|
536
|
}
|
537
|
//addExtra(jg, "used by", Joiner.on(", ").join(resourceReader.getUsedBy(res)));
|
538
|
addExtra(jg, "creation time", resourceReader.getFirstCreationTime(res));
|
539
|
|
540
|
jg.writeEndArray();
|
541
|
|
542
|
jg.writeEndObject();
|
543
|
jg.close();
|
544
|
return out.toString("UTF-8");
|
545
|
}
|
546
|
|
547
|
protected String getJsonForCollection(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
548
|
throws IOException, ParthenosPublisherException {
|
549
|
JsonFactory jsonFactory = new JsonFactory();
|
550
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
551
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
552
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
553
|
jg.writeStartObject();
|
554
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
555
|
|
556
|
jg.writeArrayFieldStart("extras");
|
557
|
addExtra(jg, "system:type", CKANUtils.Collection_type);
|
558
|
//specific class
|
559
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E70_Thing).getLocalName());
|
560
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {
|
561
|
addExtra(jg, "Parthenos URL", res.getURI());
|
562
|
}
|
563
|
else addExtra(jg, "URL", res.getURI());
|
564
|
addIdentifiers(res, jg);
|
565
|
addExtra(jg, "has type", Joiner.on(", ").join(resourceReader.getHasTypeLabels(res)));
|
566
|
addExtra(jg, "subject", Joiner.on(", ").join(resourceReader.getSubjects(res)));
|
567
|
addExtra(jg, "temporal coverage", Joiner.on(", ").join(resourceReader.getTemporalCoverages(res)));
|
568
|
addExtra(jg, "spatial coverage", Joiner.on(", ").join(resourceReader.getSpatialCoverages(res)));
|
569
|
int idx = 1;
|
570
|
Iterator<String> hasPart = resourceReader.getHasPartUrls(res);
|
571
|
while(hasPart.hasNext()){
|
572
|
addExtra(jg, String.format("has part (%d)", idx), hasPart.next());
|
573
|
idx++;
|
574
|
}
|
575
|
//addExtra(jg, "has part", Joiner.on(", ").join(resourceReader.getHasPartUrls(res)));
|
576
|
idx = 1;
|
577
|
Iterator<String> hosted = resourceReader.getHostedBys(res);
|
578
|
while(hosted.hasNext()){
|
579
|
addExtra(jg, String.format("hosted by (%d)", idx), hosted.next());
|
580
|
idx++;
|
581
|
}
|
582
|
//addExtra(jg, "hosted by", Joiner.on(", ").join(resourceReader.getHostedBys(res)));
|
583
|
idx = 1;
|
584
|
Iterator<String> curated = resourceReader.getCuratorUrls(res);
|
585
|
while(curated.hasNext()){
|
586
|
addExtra(jg, String.format("curated by (%d)", idx), curated.next());
|
587
|
idx++;
|
588
|
}
|
589
|
//addExtra(jg, "curated by", Joiner.on(", ").join(resourceReader.getCuratorUrls(res)));
|
590
|
idx = 1;
|
591
|
Iterator<String> creators = resourceReader.getCreatorsURIs(res);
|
592
|
while(creators.hasNext()){
|
593
|
addExtra(jg, String.format("creator (%d)", idx), creators.next());
|
594
|
idx++;
|
595
|
}
|
596
|
//addExtra(jg, "creator", Joiner.on(", ").join(resourceReader.getCreatorsURIs(res)));
|
597
|
addExtra(jg, "languages", Joiner.on(", ").join(resourceReader.getLanguages(res)));
|
598
|
|
599
|
jg.writeEndArray();
|
600
|
|
601
|
jg.writeEndObject();
|
602
|
jg.close();
|
603
|
return out.toString("UTF-8");
|
604
|
}
|
605
|
|
606
|
|
607
|
protected String getJsonForDesignProcedure(final Resource res, final String resNameForCatalogue, final String datasourceName)
|
608
|
throws IOException, ParthenosPublisherException {
|
609
|
JsonFactory jsonFactory = new JsonFactory();
|
610
|
final ByteArrayOutputStream out = new ByteArrayOutputStream();
|
611
|
BufferedOutputStream bos = new BufferedOutputStream(out);
|
612
|
JsonGenerator jg = jsonFactory.createGenerator(bos, JsonEncoding.UTF8);
|
613
|
jg.writeStartObject();
|
614
|
writeCommonFields(jg, res, resNameForCatalogue, datasourceName);
|
615
|
|
616
|
jg.writeArrayFieldStart("extras");
|
617
|
addExtra(jg, "system:type", CKANUtils.DesignOrProcedure_type);
|
618
|
//specific class
|
619
|
addExtra(jg, "instance of", resourceReader.findSpecificType(res, CRM.E29_Design_or_Procedure).getLocalName());
|
620
|
if (res.getURI().startsWith(PARTHENOS_BASE_URL)) {
|
621
|
addExtra(jg, "Parthenos URL", res.getURI());
|
622
|
}
|
623
|
else addExtra(jg, "URL", res.getURI());
|
624
|
addIdentifiers(res, jg);
|
625
|
int idx = 1;
|
626
|
Iterator<String> used = resourceReader.getUsedBy(res);
|
627
|
while(used.hasNext()){
|
628
|
addExtra(jg, String.format("used by (%d)", idx), used.next());
|
629
|
idx++;
|
630
|
}
|
631
|
//addExtra(jg, "used by", Joiner.on(", ").join(resourceReader.getUsedBy(res)));
|
632
|
//TODO: add additional metadata for E29_Design_or_Procedure, if any
|
633
|
jg.writeEndArray();
|
634
|
|
635
|
jg.writeEndObject();
|
636
|
jg.close();
|
637
|
return out.toString("UTF-8");
|
638
|
}
|
639
|
|
640
|
protected void addIdentifiers(final Resource res, final JsonGenerator jg ) throws IOException {
|
641
|
/*
|
642
|
<${subjectURL}> crm:P1_is_identified_by ?IDRes .
|
643
|
?IDRes a crm:E42_Identifier .
|
644
|
?IDRes rdfs:label ?ID_label .
|
645
|
*/
|
646
|
StmtIterator it = res.listProperties(CRM.P1_is_identified_by);
|
647
|
Set<String> ids = Sets.newHashSet();
|
648
|
while(it.hasNext()){
|
649
|
RDFNode obj = it.next().getObject();
|
650
|
if(obj.isLiteral()) ids.add(obj.asLiteral().getLexicalForm());
|
651
|
else {
|
652
|
Resource id = (Resource) obj;
|
653
|
if (id.hasProperty(RDF.type, CRM.E42_Identifier)) {
|
654
|
ids.add(resourceReader.getLabel(id));
|
655
|
}
|
656
|
}
|
657
|
}
|
658
|
addExtra(jg, "ID", String.join(",", ids));
|
659
|
}
|
660
|
|
661
|
protected void addExtra(final JsonGenerator jg, final String key, final String value) throws IOException {
|
662
|
if(StringUtils.isNotBlank(value)) {
|
663
|
jg.writeStartObject();
|
664
|
jg.writeStringField("key", key);
|
665
|
jg.writeStringField("value", value);
|
666
|
jg.writeEndObject();
|
667
|
}
|
668
|
}
|
669
|
|
670
|
|
671
|
|
672
|
protected void writeCommonFields(final JsonGenerator jg, final Resource res, final String resNameForCatalogue, final String datasourceName)
|
673
|
throws IOException, ParthenosPublisherException {
|
674
|
String ckanOrg = CKANUtils.getCKanOrg(datasourceName);
|
675
|
//id is available only for updates
|
676
|
if(StringUtils.isBlank(ckanOrg)) throw new ParthenosPublisherException(String.format("Cannot register %s : blank ckan org for data source with name %s", resNameForCatalogue, datasourceName));
|
677
|
//the owning organization, i.e. the data souce from which this resource has been collected from
|
678
|
jg.writeStringField("owner_org", ckanOrg);
|
679
|
jg.writeStringField("name", resNameForCatalogue);
|
680
|
//default license
|
681
|
jg.writeStringField("license_id", resourceReader.getCatalogueLicense(res).getId());
|
682
|
String title = resourceReader.getTitle(res);
|
683
|
if (StringUtils.isBlank(title))
|
684
|
title = resNameForCatalogue;
|
685
|
jg.writeStringField("title", title);
|
686
|
//description
|
687
|
jg.writeStringField("notes",Joiner.on(';').join(resourceReader.getDescriptions(res)));
|
688
|
//the names of all superclasses of the entity
|
689
|
jg.writeArrayFieldStart("tags");
|
690
|
Iterator<String> classNames = resourceReader.getRDFClassNames(res);
|
691
|
while (classNames.hasNext()) {
|
692
|
jg.writeStartObject();
|
693
|
jg.writeStringField("name", classNames.next());
|
694
|
jg.writeEndObject();
|
695
|
}
|
696
|
jg.writeEndArray();
|
697
|
}
|
698
|
|
699
|
|
700
|
public ResourceReader getResourceReader() {
|
701
|
return resourceReader;
|
702
|
}
|
703
|
|
704
|
public void setResourceReader(final ResourceReader resourceReader) {
|
705
|
this.resourceReader = resourceReader;
|
706
|
}
|
707
|
|
708
|
public CatalogueAPIClient getCatalogueAPIClient() {
|
709
|
return catalogueAPIClient;
|
710
|
}
|
711
|
|
712
|
public void setCatalogueAPIClient(final CatalogueAPIClient catalogueAPIClient) {
|
713
|
this.catalogueAPIClient = catalogueAPIClient;
|
714
|
}
|
715
|
}
|