1
|
package eu.dnetlib.data.transform.xml;
|
2
|
|
3
|
import java.util.List;
|
4
|
import java.util.Map;
|
5
|
|
6
|
import com.google.common.collect.Iterables;
|
7
|
import com.google.common.collect.Lists;
|
8
|
import com.google.protobuf.Descriptors.Descriptor;
|
9
|
import eu.dnetlib.data.mapreduce.util.OafRowKeyDecoder;
|
10
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
11
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
12
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
13
|
import eu.dnetlib.data.proto.PersonProtos.Person;
|
14
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
15
|
import eu.dnetlib.data.proto.ResultProtos.Result.Context;
|
16
|
import eu.dnetlib.data.proto.ResultProtos.Result.ExternalReference;
|
17
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
18
|
import eu.dnetlib.data.proto.ResultProtos.Result.Journal;
|
19
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
20
|
import org.apache.commons.lang.StringUtils;
|
21
|
import org.w3c.dom.NodeList;
|
22
|
|
23
|
public class OafToHbaseXsltFunctions extends CommonDNetXsltFunctions {
|
24
|
|
25
|
public static String oafResult(
|
26
|
final String resultId,
|
27
|
final String provenance,
|
28
|
final String trust,
|
29
|
final NodeList about,
|
30
|
final String hostedbyId,
|
31
|
final String hostedbyName,
|
32
|
final String collectedFromId,
|
33
|
final String collectedFromName,
|
34
|
final String originalId,
|
35
|
final String dateOfCollection,
|
36
|
final String dateOfTransformation,
|
37
|
final NodeList nodelist) {
|
38
|
try {
|
39
|
final String entityId = OafRowKeyDecoder.decode(resultId).getKey();
|
40
|
|
41
|
final Result.Builder result = Result.newBuilder();
|
42
|
|
43
|
final ValueMap values = ValueMap.parseNodeList(nodelist);
|
44
|
|
45
|
final Result.Metadata.Builder metadata = Result.Metadata.newBuilder();
|
46
|
final Descriptor mDesc = Result.Metadata.getDescriptor();
|
47
|
|
48
|
if (values.get("creator") != null) {
|
49
|
for (final String fullname : Iterables.limit(values.get("creator").listValues(), 10)) {
|
50
|
|
51
|
final Person.Metadata.Builder authorMetadata = Person.Metadata.newBuilder();
|
52
|
|
53
|
authorMetadata.setFullname(sf(fullname));
|
54
|
|
55
|
final eu.dnetlib.pace.model.Person p = new eu.dnetlib.pace.model.Person(fullname, false);
|
56
|
if (p.isAccurate()) {
|
57
|
authorMetadata.setFirstname(sf(p.getNormalisedFirstName()));
|
58
|
authorMetadata.clearSecondnames().addSecondnames(sf(p.getNormalisedSurname()));
|
59
|
authorMetadata.setFullname(sf(p.getNormalisedFullname()));
|
60
|
}
|
61
|
|
62
|
result.addAuthor(Person.newBuilder().setMetadata(authorMetadata));
|
63
|
}
|
64
|
}
|
65
|
|
66
|
addStructuredProps(metadata, mDesc.findFieldByName("subject"), values.get("subject").listValues(), "keyword", "dnet:result_subject");
|
67
|
addStructuredProps(metadata, mDesc.findFieldByName("title"), values.get("title").listValues(), "main title", "dnet:dataCite_title");
|
68
|
|
69
|
for (final String fieldname : Lists.newArrayList("description", "source")) {
|
70
|
if (values.get(fieldname) != null) {
|
71
|
for (final String s : values.get(fieldname).listValues()) {
|
72
|
addField(metadata, mDesc.findFieldByName(fieldname), s);
|
73
|
}
|
74
|
}
|
75
|
}
|
76
|
|
77
|
addField(metadata, mDesc.findFieldByName("language"), setQualifier(getDefaultQualifier("dnet:languages"), values.get("language").listValues()));
|
78
|
addField(metadata, mDesc.findFieldByName("dateofacceptance"), values.get("dateaccepted").listValues());
|
79
|
addField(metadata, mDesc.findFieldByName("publisher"), values.get("publisher").listValues());
|
80
|
addField(metadata, mDesc.findFieldByName("embargoenddate"), values.get("embargoenddate").listValues());
|
81
|
addField(metadata, mDesc.findFieldByName("storagedate"), values.get("storagedate").listValues());
|
82
|
|
83
|
addField(metadata, mDesc.findFieldByName("resulttype"), getSimpleQualifier("publication", "dnet:result_typologies"));
|
84
|
|
85
|
addField(metadata, mDesc.findFieldByName("fulltext"), values.get("fulltext").listValues());
|
86
|
addField(metadata, mDesc.findFieldByName("format"), values.get("format").listValues());
|
87
|
|
88
|
// addField(metadata, Result.Metadata.getDescriptor().findFieldByName("provenanceaction"),
|
89
|
// getSimpleQualifier("sysimport:crosswalk:repository", "dnet:provenanceActions").build());
|
90
|
|
91
|
if (values.get("concept") != null) {
|
92
|
for (final Element e : values.get("concept")) {
|
93
|
final String id = e.getAttributes().get("id");
|
94
|
if (StringUtils.isBlank(id)) throw new IllegalArgumentException("Context id cannot be blank");
|
95
|
metadata.addContext(Context.newBuilder().setId(id));
|
96
|
}
|
97
|
}
|
98
|
|
99
|
if (values.get("journal") != null) {
|
100
|
for (final Element e : values.get("journal")) {
|
101
|
|
102
|
final Journal.Builder journal = Journal.newBuilder();
|
103
|
if (e.getText() != null) {
|
104
|
journal.setName(e.getText());
|
105
|
}
|
106
|
|
107
|
final Map<String, String> attr = e.getAttributes();
|
108
|
if (attr != null) {
|
109
|
if (attr.get("issn") != null) {
|
110
|
journal.setIssnPrinted(attr.get("issn"));
|
111
|
}
|
112
|
if (attr.get("eissn") != null) {
|
113
|
journal.setIssnOnline(attr.get("eissn"));
|
114
|
}
|
115
|
if (attr.get("lissn") != null) {
|
116
|
journal.setIssnLinking(attr.get("lissn"));
|
117
|
}
|
118
|
}
|
119
|
metadata.setJournal(journal.build());
|
120
|
}
|
121
|
}
|
122
|
|
123
|
final Instance.Builder instance = Instance.newBuilder().setHostedby(getKV(hostedbyId, hostedbyName));
|
124
|
|
125
|
addField(instance, Instance.getDescriptor().findFieldByName("licence"),
|
126
|
setQualifier(getDefaultQualifier("dnet:access_modes"), values.get("accessrights").listValues()));
|
127
|
addField(instance, Instance.getDescriptor().findFieldByName("instancetype"),
|
128
|
setQualifier(getDefaultQualifier("dnet:publication_resource"), values.get("cobjcategory").listValues()));
|
129
|
|
130
|
if (values.get("identifier") != null) {
|
131
|
addField(instance, Instance.getDescriptor().findFieldByName("url"),
|
132
|
Lists.newArrayList(Iterables.filter(values.get("identifier").listValues(), urlFilter)));
|
133
|
}
|
134
|
|
135
|
result.addInstance(instance);
|
136
|
|
137
|
final List<Element> extrefs = values.get("reference");
|
138
|
if (!extrefs.isEmpty()) {
|
139
|
final Descriptor extDesc = ExternalReference.getDescriptor();
|
140
|
for (final Element element : extrefs) {
|
141
|
final ExternalReference.Builder extref = ExternalReference.newBuilder();
|
142
|
addField(extref, extDesc.findFieldByName("url"), element.getText());
|
143
|
addField(extref, extDesc.findFieldByName("sitename"), element.getAttributes().get("source"));
|
144
|
addField(extref, extDesc.findFieldByName("refidentifier"), element.getAttributes().get("identifier"));
|
145
|
addField(extref, extDesc.findFieldByName("label"), element.getAttributes().get("title"));
|
146
|
addField(extref, extDesc.findFieldByName("query"), element.getAttributes().get("query"));
|
147
|
addField(extref, extDesc.findFieldByName("qualifier"),
|
148
|
setQualifier(getDefaultQualifier("dnet:externalReference_typologies"), Lists.newArrayList(element.getAttributes().get("type")))
|
149
|
.build());
|
150
|
|
151
|
result.addExternalReference(extref);
|
152
|
}
|
153
|
}
|
154
|
|
155
|
final List<StructuredProperty> pids = Lists.newArrayList();
|
156
|
pids.addAll(parsePids(nodelist));
|
157
|
pids.add(getStructuredProperty(originalId, "oai", getClassName("oai"), "dnet:pid_types", "dnet:pid_types"));
|
158
|
|
159
|
final OafEntity.Builder entity =
|
160
|
getEntity(Type.result, entityId, getKV(collectedFromId, collectedFromName), Lists.newArrayList(originalId), dateOfCollection,
|
161
|
dateOfTransformation, pids)
|
162
|
.setResult(result.setMetadata(metadata));
|
163
|
|
164
|
entity.setOaiprovenance(getOAIProvenance(about));
|
165
|
|
166
|
final Oaf oaf = getOaf(entity, getDataInfo(about, provenance, trust, false, false));
|
167
|
return base64(oaf.toByteArray());
|
168
|
} catch (final Throwable e) {
|
169
|
System.err.println("resultId: " + resultId);
|
170
|
System.err.println("hostedbyId: " + hostedbyId);
|
171
|
System.err.println("hostedbyName: " + hostedbyName);
|
172
|
System.err.println("provenance: " + provenance);
|
173
|
System.err.println("trust: " + trust);
|
174
|
System.err.println("collectedFromId: " + collectedFromId);
|
175
|
System.err.println("collectedFromName: " + collectedFromName);
|
176
|
System.err.println("originalId: " + originalId);
|
177
|
System.err.println("dateOfCollection: " + dateOfCollection);
|
178
|
e.printStackTrace();
|
179
|
throw new RuntimeException(e);
|
180
|
}
|
181
|
}
|
182
|
|
183
|
}
|