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.google.common.collect.Lists;
|
8
|
import com.ximpleware.AutoPilot;
|
9
|
import com.ximpleware.VTDGen;
|
10
|
import com.ximpleware.VTDNav;
|
11
|
|
12
|
import eu.dnetlib.dli.resolver.model.*;
|
13
|
import eu.dnetlib.resolver.parser.UtilityParser.Node;
|
14
|
import org.apache.commons.logging.Log;
|
15
|
import org.apache.commons.logging.LogFactory;
|
16
|
|
17
|
/**
|
18
|
* Created by sandro on 9/13/16.
|
19
|
*/
|
20
|
public class DMFResolverParser extends AbstractResolverParser {
|
21
|
|
22
|
private static final Log log = LogFactory.getLog(DMFResolverParser.class);
|
23
|
|
24
|
@Override
|
25
|
public ResolvedObject parseObject(final String record) {
|
26
|
try {
|
27
|
final ResolvedObject parsedObject = new ResolvedObject();
|
28
|
final VTDGen vg = new VTDGen();
|
29
|
vg.setDoc(record.getBytes());
|
30
|
vg.parse(true);
|
31
|
|
32
|
final VTDNav vn = vg.getNav();
|
33
|
final AutoPilot ap = new AutoPilot(vn);
|
34
|
ap.declareXPathNameSpace("dri", "http://www.driver-repository.eu/namespace/dri");
|
35
|
|
36
|
final String datasourcePrefix = UtilityParser.getSingleValue(ap, vn, "//dri:datasourceprefix");
|
37
|
ap.declareXPathNameSpace("oaf", "http://namespace.dnet.eu/oaf");
|
38
|
|
39
|
final String completionStatus = UtilityParser.getSingleValue(ap, vn, "//oaf:completionStatus");
|
40
|
final String provisionMode = UtilityParser.getSingleValue(ap, vn, "//oaf:provisionMode");
|
41
|
|
42
|
final ObjectProvenance provenance = new ObjectProvenance();
|
43
|
provenance.setDatasourceId(datasourcePrefix);
|
44
|
provenance.setDatasource(datasourcePrefix);
|
45
|
provenance.setCompletionStatus(completionStatus);
|
46
|
provenance.setProvisionMode(provisionMode);
|
47
|
parsedObject.setDatasourceProvenance(Lists.newArrayList(provenance));
|
48
|
|
49
|
ap.declareXPathNameSpace("datacite", "http://datacite.org/schema/kernel-3");
|
50
|
final List<Node> identifierType =
|
51
|
UtilityParser.getTextValuesWithAttributes(ap, vn, "//datacite:identifier", Lists.newArrayList("identifierType"));
|
52
|
|
53
|
if (identifierType != null && identifierType.size() > 0) {
|
54
|
|
55
|
final Node result = identifierType.get(0);
|
56
|
parsedObject.setPid(result.getTextValue());
|
57
|
parsedObject.setPidType(result.getAttributes().get("identifierType"));
|
58
|
} else {
|
59
|
log.debug("Error on parsing record the identifier should not null ");
|
60
|
return null;
|
61
|
}
|
62
|
|
63
|
final List<Node> relations =
|
64
|
UtilityParser.getTextValuesWithAttributes(ap, vn, "//datacite:relatedIdentifier", Arrays.asList("relatedIdentifierType", "relationType"));
|
65
|
|
66
|
if (relations != null && relations.size() > 0) {
|
67
|
final List<ObjectRelation> relationsResult = new ArrayList<>();
|
68
|
relations.forEach(relationMap -> {
|
69
|
final String relationType = relationMap.getAttributes().get("relationType");
|
70
|
final String inverseRelationType = relationMap.getAttributes().get("inverseRelationType");
|
71
|
final String relatedIdentifierType = relationMap.getAttributes().get("relatedIdentifierType");
|
72
|
final String relatedPid = relationMap.getTextValue();
|
73
|
final ObjectRelation currentRelation = new ObjectRelation();
|
74
|
currentRelation.setTargetPID(new PID(relatedPid, relatedIdentifierType));
|
75
|
currentRelation.setRelationSemantics(relationType);
|
76
|
currentRelation.setInverseRelation(inverseRelationType);
|
77
|
currentRelation.setCompletionStatus(CompletionStatus.incomplete.toString());
|
78
|
relationsResult.add(currentRelation);
|
79
|
});
|
80
|
parsedObject.setRelations(relationsResult);
|
81
|
}
|
82
|
|
83
|
final List<Node> subjects = UtilityParser.getTextValuesWithAttributes(ap, vn, "//datacite:subject", Arrays.asList("subjectScheme"));
|
84
|
|
85
|
if (subjects != null && subjects.size() > 0) {
|
86
|
final List<SubjectType> subjectResult = new ArrayList<>();
|
87
|
subjects.forEach(subjectMap -> {
|
88
|
final SubjectType subject = new SubjectType(subjectMap.getAttributes().get("subjectScheme"), subjectMap.getTextValue());
|
89
|
subjectResult.add(subject);
|
90
|
});
|
91
|
parsedObject.setSubjects(subjectResult);
|
92
|
}
|
93
|
|
94
|
parsedObject.setCompletionStatus(completionStatus);
|
95
|
|
96
|
final List<String> creators = UtilityParser.getTextValue(ap, vn, "//datacite:creator/datacite:creatorName");
|
97
|
if (creators != null && creators.size() > 0) {
|
98
|
parsedObject.setAuthors(creators);
|
99
|
}
|
100
|
final List<String> titles = UtilityParser.getTextValue(ap, vn, "//datacite:title");
|
101
|
if (titles != null && titles.size() > 0) {
|
102
|
parsedObject.setTitles(titles);
|
103
|
}
|
104
|
final String type = UtilityParser.getSingleValue(ap, vn, "//datacite:resourceType");
|
105
|
|
106
|
setType(parsedObject, type);
|
107
|
|
108
|
final List<String> dates = UtilityParser.getTextValue(ap, vn, "//datacite:dates/date");
|
109
|
|
110
|
if (dates != null && dates.size() > 0) {
|
111
|
parsedObject.setDate(dates.get(0));
|
112
|
}
|
113
|
return parsedObject;
|
114
|
} catch (Throwable e) {
|
115
|
log.error("Error on parsing record " + record, e);
|
116
|
return null;
|
117
|
}
|
118
|
}
|
119
|
|
120
|
|
121
|
|
122
|
}
|
123
|
|