|
1
|
package eu.dnetlib.data.mapreduce.hbase.broker;
|
|
2
|
|
|
3
|
import static eu.dnetlib.data.mapreduce.hbase.broker.mapping.EventFactory.asEvent;
|
|
4
|
|
|
5
|
import java.io.IOException;
|
|
6
|
import java.util.List;
|
|
7
|
import java.util.stream.Collectors;
|
|
8
|
|
|
9
|
import org.apache.commons.lang.StringUtils;
|
|
10
|
import org.dom4j.DocumentException;
|
|
11
|
|
|
12
|
import com.google.common.collect.Lists;
|
|
13
|
|
|
14
|
import eu.dnetlib.broker.objects.OpenAireEventPayload;
|
|
15
|
import eu.dnetlib.data.mapreduce.hbase.broker.mapping.HighlightFactory;
|
|
16
|
import eu.dnetlib.data.mapreduce.hbase.broker.mapping.OpenAireEventPayloadFactory;
|
|
17
|
import eu.dnetlib.data.mapreduce.hbase.broker.model.EventWrapper;
|
|
18
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
|
19
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
|
20
|
|
|
21
|
/**
|
|
22
|
* Created by claudio on 26/07/16.
|
|
23
|
*/
|
|
24
|
public class AbstractEventFactory { // Abstract of the publication
|
|
25
|
|
|
26
|
public static List<EventWrapper> process(final Oaf current, final Oaf other, final float trust)
|
|
27
|
throws IOException, InterruptedException, DocumentException {
|
|
28
|
return new AbstractEventFactory().processAbstract(current, other, trust);
|
|
29
|
}
|
|
30
|
|
|
31
|
private List<EventWrapper> processAbstract(final Oaf current, final Oaf other, final float trust)
|
|
32
|
throws IOException, InterruptedException, DocumentException {
|
|
33
|
|
|
34
|
final List<EventWrapper> events = Lists.newArrayList();
|
|
35
|
|
|
36
|
if (hasAbstract(other)) {
|
|
37
|
// doProcessAbstract(context, current, other, Topic.MORE_ABSTRACT);
|
|
38
|
|
|
39
|
if (!hasAbstract(current)) {
|
|
40
|
events.add(doProcessAbstract(current, other, Topic.ENRICH_MISSING_ABSTRACT, trust));
|
|
41
|
}
|
|
42
|
}
|
|
43
|
|
|
44
|
return events;
|
|
45
|
}
|
|
46
|
|
|
47
|
private EventWrapper doProcessAbstract(final Oaf current, final Oaf other, final Topic topic, final float trust)
|
|
48
|
throws IOException, InterruptedException, DocumentException {
|
|
49
|
final Oaf.Builder prototype = Oaf.newBuilder(current);
|
|
50
|
final List<StringField> descriptionList = other.getEntity().getResult().getMetadata().getDescriptionList();
|
|
51
|
prototype.getEntityBuilder().getResultBuilder().getMetadataBuilder().addAllDescription(descriptionList);
|
|
52
|
|
|
53
|
final Oaf oaf = prototype.build();
|
|
54
|
|
|
55
|
final OpenAireEventPayload payload =
|
|
56
|
HighlightFactory.highlightEnrichAbstract(OpenAireEventPayloadFactory.fromOAF(oaf.getEntity(), other.getEntity(), trust), descriptionList);
|
|
57
|
|
|
58
|
return EventWrapper.newInstance(
|
|
59
|
asEvent(oaf.getEntity(), topic, payload, other.getEntity(), trust),
|
|
60
|
payload.getHighlight().getAbstracts().stream().filter(StringUtils::isNotBlank).sorted().collect(Collectors.joining(", ")),
|
|
61
|
topic.getValue());
|
|
62
|
}
|
|
63
|
|
|
64
|
private boolean hasAbstract(final Oaf oaf) {
|
|
65
|
return oaf.getEntity().getResult().getMetadata().getDescriptionList()
|
|
66
|
.stream()
|
|
67
|
.anyMatch(s -> StringUtils.isNotBlank(s.getValue()));
|
|
68
|
}
|
|
69
|
|
|
70
|
}
|