1 |
43783
|
sandro.lab
|
package eu.dnetlib.resolver.parser;
|
2 |
|
|
|
3 |
43939
|
sandro.lab
|
import java.util.ArrayList;
|
4 |
|
|
import java.util.Arrays;
|
5 |
|
|
import java.util.List;
|
6 |
43783
|
sandro.lab
|
|
7 |
43939
|
sandro.lab
|
import com.ximpleware.AutoPilot;
|
8 |
|
|
import com.ximpleware.VTDGen;
|
9 |
|
|
import com.ximpleware.VTDNav;
|
10 |
44356
|
claudio.at
|
import eu.dnetlib.dli.resolver.model.*;
|
11 |
45451
|
sandro.lab
|
import eu.dnetlib.data.transform.VtdUtilityParser;
|
12 |
|
|
import eu.dnetlib.data.transform.VtdUtilityParser.Node;
|
13 |
49332
|
sandro.lab
|
import eu.dnetlib.pid.resolver.model.*;
|
14 |
|
|
import eu.dnetlib.pid.resolver.parser.AbstractResolverParser;
|
15 |
43783
|
sandro.lab
|
|
16 |
|
|
/**
|
17 |
|
|
* Created by sandro on 9/13/16.
|
18 |
|
|
*/
|
19 |
|
|
public class PMFResolverParser extends AbstractResolverParser {
|
20 |
|
|
|
21 |
|
|
@Override
|
22 |
49184
|
sandro.lab
|
public DLIResolvedObject parseObject(final String record) {
|
23 |
|
|
try {
|
24 |
|
|
final DLIResolvedObject parsedObject = new DLIResolvedObject();
|
25 |
|
|
final VTDGen vg = new VTDGen();
|
26 |
43939
|
sandro.lab
|
vg.setDoc(record.getBytes());
|
27 |
|
|
vg.parse(true);
|
28 |
43783
|
sandro.lab
|
|
29 |
43939
|
sandro.lab
|
final VTDNav vn = vg.getNav();
|
30 |
|
|
final AutoPilot ap = new AutoPilot(vn);
|
31 |
43783
|
sandro.lab
|
|
32 |
45835
|
sandro.lab
|
|
33 |
49580
|
sandro.lab
|
|
34 |
|
|
final String resolvedDate = VtdUtilityParser.getSingleValue(ap, vn, "//*[local-name()='resolvedDate']");
|
35 |
45835
|
sandro.lab
|
parsedObject.setResolvedDate(resolvedDate);
|
36 |
|
|
|
37 |
43939
|
sandro.lab
|
|
38 |
49580
|
sandro.lab
|
final List<Node> pid = VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//*[local-name()='pid']", Arrays.asList("type"));
|
39 |
|
|
|
40 |
49760
|
sandro.lab
|
if (extractIdentifier(parsedObject, pid, "type")) return null;
|
41 |
43939
|
sandro.lab
|
|
42 |
|
|
List<Node> collectedFromNodes =
|
43 |
49580
|
sandro.lab
|
VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//*[local-name()='collectedFrom']", Arrays.asList("name", "id", "mode", "completionStatus"));
|
44 |
43939
|
sandro.lab
|
|
45 |
52674
|
sandro.lab
|
List<Node> resolvededFromNodes =
|
46 |
|
|
VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//*[local-name()='resolvedFrom']", Arrays.asList("name", "id", "mode", "completionStatus"));
|
47 |
|
|
|
48 |
49332
|
sandro.lab
|
final List<ObjectProvenance> provenances = new ArrayList<>();
|
49 |
43939
|
sandro.lab
|
|
50 |
|
|
if (collectedFromNodes != null && collectedFromNodes.size() > 0) {
|
51 |
|
|
collectedFromNodes.forEach(it -> {
|
52 |
49184
|
sandro.lab
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
53 |
|
|
provenance.setDatasourceId(it.getAttributes().get("id"));
|
54 |
43939
|
sandro.lab
|
provenance.setDatasource(it.getAttributes().get("name"));
|
55 |
|
|
provenance.setProvisionMode(it.getAttributes().get("mode"));
|
56 |
|
|
provenance.setCompletionStatus(it.getAttributes().get("completionStatus"));
|
57 |
|
|
provenances.add(provenance);
|
58 |
|
|
});
|
59 |
|
|
}
|
60 |
|
|
|
61 |
52674
|
sandro.lab
|
if (resolvededFromNodes != null && resolvededFromNodes.size() > 0) {
|
62 |
|
|
resolvededFromNodes.forEach(it -> {
|
63 |
|
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
64 |
|
|
provenance.setDatasourceId(it.getAttributes().get("id"));
|
65 |
|
|
provenance.setDatasource(it.getAttributes().get("name"));
|
66 |
|
|
provenance.setProvisionMode("resolved");
|
67 |
|
|
provenance.setCompletionStatus(it.getAttributes().get("completionStatus"));
|
68 |
|
|
provenances.add(provenance);
|
69 |
|
|
});
|
70 |
|
|
}
|
71 |
|
|
|
72 |
43939
|
sandro.lab
|
parsedObject.setDatasourceProvenance(provenances);
|
73 |
49580
|
sandro.lab
|
parsedObject.setCompletionStatus(VtdUtilityParser.getSingleValue(ap, vn, "//*[local-name()='completionStatus']"));
|
74 |
43939
|
sandro.lab
|
|
75 |
|
|
List<Node> relatedIdentifiers =
|
76 |
49580
|
sandro.lab
|
VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//*[local-name()='relatedIdentifier']",
|
77 |
45425
|
sandro.lab
|
Arrays.asList("relatedIdentifierType", "relationType", "entityType", "inverseRelationType"));
|
78 |
43939
|
sandro.lab
|
|
79 |
|
|
if (relatedIdentifiers != null && relatedIdentifiers.size() > 0) {
|
80 |
49216
|
sandro.lab
|
final List<ObjectRelation> relations = new ArrayList<>();
|
81 |
|
|
relatedIdentifiers.forEach(relation -> {
|
82 |
44352
|
sandro.lab
|
final String currentPid = relation.getTextValue();
|
83 |
|
|
final String currentPidType = relation.getAttributes().get("relatedIdentifierType");
|
84 |
|
|
final String currentType = relation.getAttributes().get("entityType");
|
85 |
49184
|
sandro.lab
|
final DLIObjectRelation currentRelation = new DLIObjectRelation();
|
86 |
|
|
currentRelation.setTargetPID(new PID(currentPid, currentPidType));
|
87 |
45451
|
sandro.lab
|
currentRelation.setCompletionStatus(CompletionStatus.incomplete.toString());
|
88 |
|
|
currentRelation.setSourcePid(parsedObject.getPid());
|
89 |
|
|
currentRelation.setRelationSemantics(relation.getAttributes().get("relationType"));
|
90 |
|
|
currentRelation.setInverseRelation(relation.getAttributes().get("inverseRelationType"));
|
91 |
45425
|
sandro.lab
|
|
92 |
|
|
|
93 |
45451
|
sandro.lab
|
currentRelation.setTargetType(ObjectType.valueOf(currentType));
|
94 |
43939
|
sandro.lab
|
if (parsedObject.getDatasourceProvenance() != null && parsedObject.getDatasourceProvenance().size() > 0) {
|
95 |
49332
|
sandro.lab
|
final DLIObjectProvenance provenance = (DLIObjectProvenance) parsedObject.getDatasourceProvenance().get(0);
|
96 |
43939
|
sandro.lab
|
|
97 |
49184
|
sandro.lab
|
final DLIObjectProvenance newProvenance = new DLIObjectProvenance();
|
98 |
|
|
newProvenance.setCompletionStatus(CompletionStatus.incomplete.toString());
|
99 |
43939
|
sandro.lab
|
newProvenance.setDatasourceId(provenance.getDatasourceId());
|
100 |
|
|
newProvenance.setDatasource(provenance.getDatasource());
|
101 |
|
|
newProvenance.setProvisionMode(provenance.getProvisionMode());
|
102 |
45451
|
sandro.lab
|
currentRelation.setRelationProvenance(Arrays.asList(newProvenance));
|
103 |
43939
|
sandro.lab
|
}
|
104 |
|
|
|
105 |
45451
|
sandro.lab
|
relations.add(currentRelation);
|
106 |
43939
|
sandro.lab
|
});
|
107 |
|
|
parsedObject.setRelations(relations);
|
108 |
|
|
}
|
109 |
|
|
|
110 |
49580
|
sandro.lab
|
|
111 |
|
|
final List<String> authorsNode = VtdUtilityParser.getTextValue(ap, vn, "//*[local-name()='creator']");
|
112 |
43939
|
sandro.lab
|
parsedObject.setAuthors(authorsNode);
|
113 |
49580
|
sandro.lab
|
parsedObject.setTitles(VtdUtilityParser.getTextValue(ap, vn, "//*[local-name()='title']"));
|
114 |
43939
|
sandro.lab
|
|
115 |
49580
|
sandro.lab
|
parsedObject.setDescription(VtdUtilityParser.getSingleValue(ap, vn, "//*[local-name()='description']"));
|
116 |
|
|
parsedObject.setDate(VtdUtilityParser.getSingleValue(ap, vn, "//*[local-name()='date']"));
|
117 |
|
|
List<Node> subjects = VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//*[local-name()='subject']", Arrays.asList("scheme"));
|
118 |
49332
|
sandro.lab
|
extractSubject(parsedObject, subjects);
|
119 |
43939
|
sandro.lab
|
|
120 |
|
|
|
121 |
49580
|
sandro.lab
|
setType(parsedObject, VtdUtilityParser.getSingleValue(ap, vn, "//*[local-name()='type']"));
|
122 |
43939
|
sandro.lab
|
|
123 |
|
|
return parsedObject;
|
124 |
43783
|
sandro.lab
|
} catch (Throwable e) {
|
125 |
54390
|
sandro.lab
|
log.error("Input record: " + record);
|
126 |
43783
|
sandro.lab
|
log.error("Error on parsing record ", e);
|
127 |
|
|
return null;
|
128 |
|
|
}
|
129 |
|
|
|
130 |
|
|
}
|
131 |
|
|
}
|