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