1 |
48145
|
claudio.at
|
package eu.dnetlib.data.mapreduce.hbase.broker.enrich;
|
2 |
|
|
|
3 |
|
|
import java.io.IOException;
|
4 |
|
|
import java.util.Arrays;
|
5 |
|
|
import java.util.Map;
|
6 |
|
|
|
7 |
|
|
import com.google.common.base.Function;
|
8 |
|
|
import com.google.common.base.Predicate;
|
9 |
|
|
import com.google.common.collect.Iterables;
|
10 |
|
|
import com.google.common.collect.Lists;
|
11 |
|
|
import eu.dnetlib.data.mapreduce.util.DedupUtils;
|
12 |
|
|
import eu.dnetlib.data.mapreduce.util.OafDecoder;
|
13 |
|
|
import eu.dnetlib.data.mapreduce.util.OafRowKeyDecoder;
|
14 |
|
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
15 |
|
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
16 |
|
|
import org.apache.commons.collections.MapUtils;
|
17 |
|
|
import org.apache.hadoop.hbase.client.Result;
|
18 |
|
|
import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
|
19 |
|
|
import org.apache.hadoop.hbase.util.Bytes;
|
20 |
|
|
|
21 |
|
|
/**
|
22 |
|
|
* Created by claudio on 08/07/16.
|
23 |
|
|
*/
|
24 |
|
|
public class ProjectEnrichmentMapper extends AbstractEnrichmentMapper {
|
25 |
|
|
|
26 |
|
|
@Override
|
27 |
|
|
protected String counterGroup() {
|
28 |
|
|
return "Broker Enrichment projects";
|
29 |
|
|
}
|
30 |
|
|
|
31 |
|
|
@Override
|
32 |
|
|
protected void map(final ImmutableBytesWritable key, final Result value, final Context context) throws IOException, InterruptedException {
|
33 |
|
|
|
34 |
|
|
final Type type = getEntityType(key);
|
35 |
|
|
|
36 |
|
|
final byte[] body = value.getValue(Bytes.toBytes(type.toString()), Bytes.toBytes("body"));
|
37 |
|
|
|
38 |
|
|
if (body == null) {
|
39 |
|
|
context.getCounter(counterGroup(), "missing metadata").increment(1);
|
40 |
|
|
return;
|
41 |
|
|
}
|
42 |
|
|
|
43 |
|
|
switch (type) {
|
44 |
|
|
case project:
|
45 |
|
|
for (final byte[] resultId : listRelatedIds(value, "resultProject_outcome_produces")) {
|
46 |
|
|
emit(context, resultId, body, "project");
|
47 |
|
|
}
|
48 |
|
|
break;
|
49 |
|
|
case result:
|
50 |
|
|
|
51 |
|
|
final Oaf.Builder oafBuilder = Oaf.newBuilder(OafDecoder.decode(body).getOaf());
|
52 |
51451
|
claudio.at
|
for (final String relName : Arrays.asList("resultProject_outcome_isProducedBy")) { // TODO add dataset rels
|
53 |
48145
|
claudio.at
|
for (final Oaf rel : listRelations(value, relName)) {
|
54 |
|
|
oafBuilder.getEntityBuilder().addCachedOafRel(rel);
|
55 |
|
|
context.getCounter(counterGroup(), "rel: " + relName).increment(1);
|
56 |
|
|
}
|
57 |
|
|
}
|
58 |
|
|
|
59 |
|
|
final Map<byte[], byte[]> mergedIn = value.getFamilyMap(DedupUtils.getDedupCF_mergedInBytes(Type.result));
|
60 |
|
|
|
61 |
|
|
if (MapUtils.isEmpty(mergedIn) & !DedupUtils.isRoot(key)) {
|
62 |
|
|
emit(context, key.copyBytes(), oafBuilder.build().toByteArray(), "result not deduped");
|
63 |
|
|
} else if (DedupUtils.isRoot(key)) {
|
64 |
|
|
emit(context, key.copyBytes(), oafBuilder.build().toByteArray(), "result merges");
|
65 |
|
|
} else {
|
66 |
|
|
emit(context, getRootId(mergedIn), oafBuilder.build().toByteArray(), "result mergedIn");
|
67 |
|
|
}
|
68 |
|
|
|
69 |
|
|
break;
|
70 |
|
|
default:
|
71 |
|
|
throw new IllegalArgumentException("invalid type: " + type);
|
72 |
|
|
}
|
73 |
|
|
}
|
74 |
|
|
|
75 |
|
|
private Iterable<Oaf> listRelations(final Result value, final String relType) {
|
76 |
|
|
|
77 |
|
|
//TODO consider only relationshipt not deletedbyinference
|
78 |
|
|
|
79 |
|
|
final Map<byte[], byte[]> map = value.getFamilyMap(Bytes.toBytes(relType));
|
80 |
|
|
if (MapUtils.isNotEmpty(map)) {
|
81 |
|
|
return Iterables.filter(Iterables.transform(map.values(), new Function<byte[], Oaf>() {
|
82 |
|
|
|
83 |
|
|
@Override
|
84 |
|
|
public Oaf apply(final byte[] input) {
|
85 |
|
|
return OafDecoder.decode(input).getOaf();
|
86 |
|
|
}
|
87 |
|
|
}), new Predicate<Oaf>() {
|
88 |
|
|
@Override
|
89 |
|
|
public boolean apply(final Oaf rel) {
|
90 |
|
|
return !rel.getRel().getTarget().contains("unidentified");
|
91 |
|
|
}
|
92 |
|
|
});
|
93 |
|
|
} else {
|
94 |
|
|
return Lists.newArrayList();
|
95 |
|
|
}
|
96 |
|
|
}
|
97 |
|
|
|
98 |
|
|
private Iterable<byte[]> listRelatedIds(final Result value, final String relType) {
|
99 |
|
|
|
100 |
|
|
//TODO consider only relationshipt not deletedbyinference
|
101 |
|
|
|
102 |
|
|
final Map<byte[], byte[]> map = value.getFamilyMap(Bytes.toBytes(relType));
|
103 |
|
|
if (MapUtils.isNotEmpty(map)) {
|
104 |
|
|
return map.keySet();
|
105 |
|
|
} else {
|
106 |
|
|
return Lists.newArrayList();
|
107 |
|
|
}
|
108 |
|
|
}
|
109 |
|
|
|
110 |
|
|
private Type getEntityType(final ImmutableBytesWritable key) {
|
111 |
|
|
return OafRowKeyDecoder.decode(key.copyBytes()).getType();
|
112 |
|
|
}
|
113 |
|
|
|
114 |
|
|
private byte[] getRootId(final Map<byte[], byte[]> mergedIn) {
|
115 |
|
|
return Iterables.getOnlyElement(mergedIn.keySet());
|
116 |
|
|
}
|
117 |
|
|
|
118 |
|
|
}
|