1 |
50937
|
alessia.ba
|
package eu.dnetlib.parthenos.jrr;
|
2 |
|
|
|
3 |
|
|
import java.io.IOException;
|
4 |
51287
|
alessia.ba
|
import java.net.URISyntaxException;
|
5 |
51321
|
alessia.ba
|
import javax.annotation.PostConstruct;
|
6 |
50937
|
alessia.ba
|
|
7 |
|
|
import eu.dnetlib.parthenos.CRM;
|
8 |
51321
|
alessia.ba
|
import eu.dnetlib.parthenos.CRMdig;
|
9 |
50937
|
alessia.ba
|
import eu.dnetlib.parthenos.CRMpe;
|
10 |
|
|
import eu.dnetlib.parthenos.catalogue.CatalogueRegistrator;
|
11 |
|
|
import eu.dnetlib.parthenos.publisher.ParthenosPublisherException;
|
12 |
51276
|
alessia.ba
|
import eu.dnetlib.parthenos.rdf.RecordParserHelper;
|
13 |
51429
|
alessia.ba
|
import org.apache.commons.io.IOUtils;
|
14 |
50937
|
alessia.ba
|
import org.apache.commons.lang3.StringUtils;
|
15 |
|
|
import org.apache.commons.logging.Log;
|
16 |
|
|
import org.apache.commons.logging.LogFactory;
|
17 |
51321
|
alessia.ba
|
import org.apache.jena.ontology.OntModel;
|
18 |
|
|
import org.apache.jena.ontology.OntModelSpec;
|
19 |
51096
|
alessia.ba
|
import org.apache.jena.rdf.model.*;
|
20 |
50937
|
alessia.ba
|
import org.apache.jena.vocabulary.RDF;
|
21 |
|
|
import org.springframework.beans.factory.annotation.Autowired;
|
22 |
|
|
import org.springframework.stereotype.Component;
|
23 |
|
|
|
24 |
51663
|
alessia.ba
|
|
25 |
50937
|
alessia.ba
|
/**
|
26 |
|
|
* Created by Alessia Bardi on 25/02/2018.
|
27 |
|
|
*
|
28 |
|
|
* publish on Gcube registry and catalogue.
|
29 |
|
|
*
|
30 |
|
|
*
|
31 |
|
|
* @author Alessia Bardi
|
32 |
|
|
*/
|
33 |
|
|
@Component
|
34 |
|
|
public class JRRPublisher {
|
35 |
|
|
|
36 |
|
|
private static final Log log = LogFactory.getLog(JRRPublisher.class);
|
37 |
51321
|
alessia.ba
|
private OntModel baseModel;
|
38 |
50937
|
alessia.ba
|
|
39 |
|
|
@Autowired
|
40 |
|
|
private CatalogueRegistrator catalogueRegistrator;
|
41 |
55561
|
alessia.ba
|
//@Autowired
|
42 |
|
|
//private GCubeResourceRegistrator gCubeResourceRegistrator;
|
43 |
50937
|
alessia.ba
|
@Autowired
|
44 |
51276
|
alessia.ba
|
private RecordParserHelper recordParserHelper;
|
45 |
50937
|
alessia.ba
|
|
46 |
51321
|
alessia.ba
|
@PostConstruct
|
47 |
|
|
public void init(){
|
48 |
|
|
baseModel = ModelFactory.createOntologyModel(OntModelSpec.RDFS_MEM_TRANS_INF);
|
49 |
|
|
baseModel.read(CRMpe.RDFS_URL);
|
50 |
|
|
baseModel.read(CRM.RDFS_URL);
|
51 |
|
|
baseModel.read(CRMdig.RDFS_URL);
|
52 |
|
|
}
|
53 |
|
|
|
54 |
57139
|
alessia.ba
|
//No need to pass the RDF inside an OAI record: use register(String, String, String) to explicitely pass the needed information.
|
55 |
|
|
@Deprecated
|
56 |
51767
|
alessia.ba
|
public void register(final String record)
|
57 |
55561
|
alessia.ba
|
throws ParthenosPublisherException, IOException, URISyntaxException, InterruptedException {
|
58 |
51276
|
alessia.ba
|
String id = recordParserHelper.getObjIdentifier(record);
|
59 |
51709
|
alessia.ba
|
String datasourceName = recordParserHelper.getDatasourceName(record);
|
60 |
56986
|
alessia.ba
|
String rdfRecord = recordParserHelper.getRDF(record);
|
61 |
|
|
register(rdfRecord, id, datasourceName);
|
62 |
|
|
}
|
63 |
|
|
|
64 |
|
|
public void register(final String rdfRecord, final String objIdentifier, final String datasourceName)
|
65 |
|
|
throws ParthenosPublisherException, IOException, URISyntaxException, InterruptedException {
|
66 |
51696
|
alessia.ba
|
//objIdentifier contains the subject URI used to get the RDF: that is the only resource we have to register when processing this rdf file!
|
67 |
56986
|
alessia.ba
|
log.debug("REGISTERING ON JRR: "+objIdentifier);
|
68 |
51321
|
alessia.ba
|
InfModel model = loadBaseModel();
|
69 |
51429
|
alessia.ba
|
model.read(IOUtils.toInputStream(rdfRecord), CRMpe.NS, "RDF/XML");
|
70 |
56986
|
alessia.ba
|
register(model, objIdentifier, datasourceName);
|
71 |
50937
|
alessia.ba
|
}
|
72 |
|
|
|
73 |
51321
|
alessia.ba
|
|
74 |
51767
|
alessia.ba
|
protected void register(final Model model, final String resourceURI, final String datasourceName)
|
75 |
|
|
throws ParthenosPublisherException, IOException, URISyntaxException, InterruptedException {
|
76 |
51696
|
alessia.ba
|
Resource rdfResource = model.getResource(resourceURI);
|
77 |
52248
|
alessia.ba
|
if (rdfResource == null) {
|
78 |
|
|
log.error("UNEXPECTED NULL rdfResource with resourceURI " + resourceURI + ". I am skipping it, but you should check!");
|
79 |
|
|
} else {
|
80 |
51696
|
alessia.ba
|
//call the correct register method based on the resource type
|
81 |
|
|
//we skip everything that is not Software, Actor, Service, Dataset, Curation Plan and Project: other entities are in fact source of metadata for them
|
82 |
54681
|
alessia.ba
|
////base types are {Project, Service, Actors, Datasets, Software, Physical Collections, Standards} i.e. {PE35_Project, PE1_Service, E39_Actor, PE18_Dataset, D14_Software, E78_Collection, E29_Design_or_Procedure)
|
83 |
51709
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRM.E39_Actor)) registerRDFResource(rdfResource, CRM.E39_Actor, datasourceName);
|
84 |
51696
|
alessia.ba
|
else {
|
85 |
52248
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRMpe.PE35_Project)) registerRDFResource(rdfResource, CRMpe.PE35_Project, datasourceName);
|
86 |
51696
|
alessia.ba
|
else {
|
87 |
53694
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRM.E29_Design_or_Procedure)) registerRDFResource(rdfResource, CRM.E29_Design_or_Procedure, datasourceName);
|
88 |
51696
|
alessia.ba
|
else {
|
89 |
53694
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRMdig.D14_Software)) registerRDFResource(rdfResource, CRMdig.D14_Software, datasourceName);
|
90 |
52248
|
alessia.ba
|
else {
|
91 |
53694
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRMpe.PE18_Dataset)) registerRDFResource(rdfResource, CRMpe.PE18_Dataset, datasourceName);
|
92 |
52248
|
alessia.ba
|
else {
|
93 |
|
|
if (rdfResource.hasProperty(RDF.type, CRM.E78_Collection)) registerRDFResource(rdfResource, CRM.E78_Collection, datasourceName);
|
94 |
|
|
else {
|
95 |
53694
|
alessia.ba
|
if (rdfResource.hasProperty(RDF.type, CRMpe.PE1_Service)) registerRDFResource(rdfResource, CRMpe.PE1_Service, datasourceName);
|
96 |
52248
|
alessia.ba
|
else {
|
97 |
|
|
log.debug("Skipping " + resourceURI + " because of its type");
|
98 |
|
|
}
|
99 |
|
|
}
|
100 |
|
|
}
|
101 |
|
|
}
|
102 |
51696
|
alessia.ba
|
}
|
103 |
|
|
}
|
104 |
50937
|
alessia.ba
|
}
|
105 |
|
|
}
|
106 |
|
|
}
|
107 |
|
|
|
108 |
52754
|
alessia.ba
|
protected boolean registerRDFResource(final Resource rdfResource, final Resource type, final String datasourceName)
|
109 |
51767
|
alessia.ba
|
throws ParthenosPublisherException, IOException, URISyntaxException, InterruptedException {
|
110 |
50937
|
alessia.ba
|
String resURI = rdfResource.getURI();
|
111 |
|
|
if (!resURI.startsWith("http")) {
|
112 |
|
|
//this is something George said: if it has no http URI, then it is not to be considered relevant by itself
|
113 |
|
|
log.info("Resource " + resURI + " skipped: URI does not start with http");
|
114 |
|
|
return false;
|
115 |
|
|
}
|
116 |
|
|
else {
|
117 |
52754
|
alessia.ba
|
String uuid = registerOnCatalogue(rdfResource, type, datasourceName);
|
118 |
51663
|
alessia.ba
|
if(StringUtils.isNotBlank(uuid)){
|
119 |
|
|
//TODO: let's skip the registration on the registry for now.
|
120 |
|
|
//registerOnRegistry(rdfResource, uuid, type);
|
121 |
|
|
return true;
|
122 |
|
|
}
|
123 |
|
|
else{
|
124 |
|
|
log.warn("Got blank uuid when registering "+resURI+": skipping registration on the registry");
|
125 |
|
|
return false;
|
126 |
|
|
}
|
127 |
50937
|
alessia.ba
|
}
|
128 |
|
|
}
|
129 |
|
|
|
130 |
|
|
/**
|
131 |
|
|
* Register resource of the given type on the catalogue
|
132 |
|
|
* @param rdfResource
|
133 |
|
|
* @param type
|
134 |
|
|
* @return the catalogue uuid
|
135 |
|
|
* @throws IOException
|
136 |
|
|
*/
|
137 |
52754
|
alessia.ba
|
protected String registerOnCatalogue(final Resource rdfResource, final Resource type, final String datasourceName)
|
138 |
51767
|
alessia.ba
|
throws ParthenosPublisherException, IOException, URISyntaxException, InterruptedException {
|
139 |
52754
|
alessia.ba
|
return catalogueRegistrator.register(rdfResource, type, datasourceName);
|
140 |
50937
|
alessia.ba
|
}
|
141 |
|
|
|
142 |
55561
|
alessia.ba
|
/*
|
143 |
50937
|
alessia.ba
|
protected void registerOnRegistry(final Resource rdfResource, final String uuid, final Resource type)
|
144 |
|
|
throws ParthenosPublisherException, ResourceRegistryException, IOException {
|
145 |
|
|
gCubeResourceRegistrator.register(rdfResource, uuid, type);
|
146 |
51696
|
alessia.ba
|
}
|
147 |
55561
|
alessia.ba
|
*/
|
148 |
50937
|
alessia.ba
|
|
149 |
51696
|
alessia.ba
|
protected InfModel loadBaseModel() {
|
150 |
|
|
return ModelFactory.createRDFSModel(baseModel);
|
151 |
50937
|
alessia.ba
|
}
|
152 |
|
|
|
153 |
51696
|
alessia.ba
|
|
154 |
50937
|
alessia.ba
|
}
|