1
|
package eu.dnetlib.resolver.parser;
|
2
|
|
3
|
import java.util.ArrayList;
|
4
|
import java.util.Arrays;
|
5
|
import java.util.List;
|
6
|
|
7
|
import com.ximpleware.AutoPilot;
|
8
|
import com.ximpleware.VTDGen;
|
9
|
import com.ximpleware.VTDNav;
|
10
|
import eu.dnetlib.dli.resolver.model.*;
|
11
|
import eu.dnetlib.data.transform.VtdUtilityParser;
|
12
|
import eu.dnetlib.data.transform.VtdUtilityParser.Node;
|
13
|
import eu.dnetlib.pid.resolver.model.ObjectType;
|
14
|
|
15
|
/**
|
16
|
* Created by sandro on 9/13/16.
|
17
|
*/
|
18
|
public class PMFResolverParser extends AbstractResolverParser {
|
19
|
|
20
|
@Override
|
21
|
public DLIResolvedObject parseObject(final String record) {
|
22
|
try {
|
23
|
final DLIResolvedObject parsedObject = new DLIResolvedObject();
|
24
|
final VTDGen vg = new VTDGen();
|
25
|
vg.setDoc(record.getBytes());
|
26
|
vg.parse(true);
|
27
|
|
28
|
final VTDNav vn = vg.getNav();
|
29
|
final AutoPilot ap = new AutoPilot(vn);
|
30
|
|
31
|
ap.declareXPathNameSpace("dri", "http://www.driver-repository.eu/namespace/dri");
|
32
|
|
33
|
final String resolvedDate = VtdUtilityParser.getSingleValue(ap, vn, "//dri:resolvedDate");
|
34
|
parsedObject.setResolvedDate(resolvedDate);
|
35
|
|
36
|
ap.declareXPathNameSpace("oaf", "http://namespace.dnet.eu/oaf");
|
37
|
final List<Node> pid = VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//oaf:pid", Arrays.asList("type"));
|
38
|
|
39
|
if (pid != null && pid.size() > 0) {
|
40
|
final String currentPid = pid.get(0).getTextValue();
|
41
|
final String currentPidType = pid.get(0).getAttributes().get("type");
|
42
|
parsedObject.setPid(currentPid);
|
43
|
parsedObject.setPidType(currentPidType);
|
44
|
} else {
|
45
|
return null;
|
46
|
}
|
47
|
|
48
|
List<Node> collectedFromNodes =
|
49
|
VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//oaf:collectedFrom", Arrays.asList("name", "id", "mode", "completionStatus"));
|
50
|
|
51
|
final List<DLIObjectProvenance> provenances = new ArrayList<>();
|
52
|
|
53
|
if (collectedFromNodes != null && collectedFromNodes.size() > 0) {
|
54
|
collectedFromNodes.forEach(it -> {
|
55
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
56
|
provenance.setDatasourceId(it.getAttributes().get("id"));
|
57
|
provenance.setDatasource(it.getAttributes().get("name"));
|
58
|
provenance.setProvisionMode(it.getAttributes().get("mode"));
|
59
|
provenance.setCompletionStatus(it.getAttributes().get("completionStatus"));
|
60
|
provenances.add(provenance);
|
61
|
});
|
62
|
}
|
63
|
|
64
|
parsedObject.setDatasourceProvenance(provenances);
|
65
|
parsedObject.setCompletionStatus(VtdUtilityParser.getSingleValue(ap, vn, "//oaf:completionStatus"));
|
66
|
|
67
|
List<Node> relatedIdentifiers =
|
68
|
VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//oaf:relatedIdentifier",
|
69
|
Arrays.asList("relatedIdentifierType", "relationType", "entityType", "inverseRelationType"));
|
70
|
|
71
|
if (relatedIdentifiers != null && relatedIdentifiers.size() > 0) {
|
72
|
final List<DLIObjectRelation> relations = new ArrayList<>();
|
73
|
relatedIdentifiers.forEach(relation -> {
|
74
|
final String currentPid = relation.getTextValue();
|
75
|
final String currentPidType = relation.getAttributes().get("relatedIdentifierType");
|
76
|
final String currentType = relation.getAttributes().get("entityType");
|
77
|
final DLIObjectRelation currentRelation = new DLIObjectRelation();
|
78
|
currentRelation.setTargetPID(new PID(currentPid, currentPidType));
|
79
|
currentRelation.setCompletionStatus(CompletionStatus.incomplete.toString());
|
80
|
currentRelation.setSourcePid(parsedObject.getPid());
|
81
|
currentRelation.setRelationSemantics(relation.getAttributes().get("relationType"));
|
82
|
currentRelation.setInverseRelation(relation.getAttributes().get("inverseRelationType"));
|
83
|
|
84
|
|
85
|
currentRelation.setTargetType(ObjectType.valueOf(currentType));
|
86
|
if (parsedObject.getDatasourceProvenance() != null && parsedObject.getDatasourceProvenance().size() > 0) {
|
87
|
final DLIObjectProvenance provenance = parsedObject.getDatasourceProvenance().get(0);
|
88
|
|
89
|
final DLIObjectProvenance newProvenance = new DLIObjectProvenance();
|
90
|
newProvenance.setCompletionStatus(CompletionStatus.incomplete.toString());
|
91
|
newProvenance.setDatasourceId(provenance.getDatasourceId());
|
92
|
newProvenance.setDatasource(provenance.getDatasource());
|
93
|
newProvenance.setProvisionMode(provenance.getProvisionMode());
|
94
|
currentRelation.setRelationProvenance(Arrays.asList(newProvenance));
|
95
|
}
|
96
|
|
97
|
relations.add(currentRelation);
|
98
|
});
|
99
|
parsedObject.setRelations(relations);
|
100
|
}
|
101
|
|
102
|
ap.declareXPathNameSpace("dc", "http://purl.org/dc/elements/1.1/");
|
103
|
final List<String> authorsNode = VtdUtilityParser.getTextValue(ap, vn, "//dc:creator");
|
104
|
parsedObject.setAuthors(authorsNode);
|
105
|
parsedObject.setTitles(VtdUtilityParser.getTextValue(ap, vn, "//dc:title"));
|
106
|
|
107
|
parsedObject.setDescription(VtdUtilityParser.getSingleValue(ap, vn, "//dc:description"));
|
108
|
parsedObject.setDate(VtdUtilityParser.getSingleValue(ap, vn, "//dc:date"));
|
109
|
List<Node> subjects = VtdUtilityParser.getTextValuesWithAttributes(ap, vn, "//dc:subject", Arrays.asList("scheme"));
|
110
|
if (subjects != null && subjects.size() > 0) {
|
111
|
final List<SubjectType> currentSubjects = new ArrayList<>();
|
112
|
subjects.forEach(it -> {
|
113
|
|
114
|
String scheme = it.getAttributes().get("scheme");
|
115
|
if (scheme == null)
|
116
|
scheme = "unknown";
|
117
|
currentSubjects.add(new SubjectType(scheme, it.getTextValue()));
|
118
|
});
|
119
|
|
120
|
parsedObject.setSubjects(currentSubjects);
|
121
|
}
|
122
|
|
123
|
setType(parsedObject, VtdUtilityParser.getSingleValue(ap, vn, "//dc:type"));
|
124
|
|
125
|
return parsedObject;
|
126
|
} catch (Throwable e) {
|
127
|
log.debug("Input record: " + record);
|
128
|
log.error("Error on parsing record ", e);
|
129
|
return null;
|
130
|
}
|
131
|
|
132
|
}
|
133
|
}
|