1
|
package eu.dnetlib.data.mapreduce.hbase.dedup;
|
2
|
|
3
|
import java.io.IOException;
|
4
|
import java.util.List;
|
5
|
|
6
|
import eu.dnetlib.data.graph.utils.RelDescriptor;
|
7
|
import eu.dnetlib.data.mapreduce.JobParams;
|
8
|
import eu.dnetlib.data.proto.DNGFProtos.DNGF;
|
9
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
10
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
11
|
import eu.dnetlib.pace.config.DedupConfig;
|
12
|
import org.apache.commons.logging.Log;
|
13
|
import org.apache.commons.logging.LogFactory;
|
14
|
import org.apache.hadoop.hbase.client.Result;
|
15
|
import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
|
16
|
import org.apache.hadoop.hbase.mapreduce.TableMapper;
|
17
|
import org.apache.hadoop.io.Text;
|
18
|
|
19
|
import static eu.dnetlib.data.mapreduce.util.UpdateMerger.mergeBodyUpdates;
|
20
|
import static eu.dnetlib.data.mapreduce.util.dao.HBaseTableDAO.*;
|
21
|
|
22
|
public class DedupBuildRootsMapper extends TableMapper<Text, ImmutableBytesWritable> {
|
23
|
|
24
|
private static final Log log = LogFactory.getLog(DedupBuildRootsMapper.class);
|
25
|
|
26
|
private DedupConfig dedupConf;
|
27
|
|
28
|
private ImmutableBytesWritable ibw;
|
29
|
|
30
|
@Override
|
31
|
protected void setup(final Context context) {
|
32
|
dedupConf = DedupConfig.load(context.getConfiguration().get(JobParams.DEDUP_CONF));
|
33
|
System.out.println("dedup buildRoots mapper\nwf conf: " + dedupConf.toString());
|
34
|
|
35
|
ibw = new ImmutableBytesWritable();
|
36
|
}
|
37
|
|
38
|
@Override
|
39
|
protected void map(final ImmutableBytesWritable rowkey, final Result value, final Context context) throws IOException, InterruptedException {
|
40
|
// System.out.println("Find root mapping: " + new String(rowkey.copyBytes()));
|
41
|
|
42
|
final Type type = Type.valueOf(dedupConf.getWf().getEntityType());
|
43
|
final List<String> mergedIn = getTargetIds(value, "isMergedIn");
|
44
|
|
45
|
if ((mergedIn != null) && !mergedIn.isEmpty()) {
|
46
|
|
47
|
if (mergedIn.size() > 1) {
|
48
|
throw new RuntimeException("found more than one merged in relation for row key: " + new String(rowkey.copyBytes()));
|
49
|
}
|
50
|
final Text rootId = new Text(mergedIn.get(0));
|
51
|
|
52
|
context.getCounter(dedupConf.getWf().getEntityType(), "merged").increment(1);
|
53
|
|
54
|
final DNGF dngfMerged = mergeBodyUpdates(context, value.getFamilyMap(cfMetadataByte()), type);
|
55
|
if (dngfMerged == null) {
|
56
|
context.getCounter(dedupConf.getWf().getEntityType(), "missing body").increment(1);
|
57
|
System.out.println("missing body in: " + new String(rowkey.copyBytes()));
|
58
|
return;
|
59
|
}
|
60
|
emit(context, rootId, dngfMerged.toByteArray());
|
61
|
|
62
|
rel(value, "isMergedIn", "merges", "isSimilarTo").values().forEach(dngf -> {
|
63
|
if (!isRelMarkedDeleted(dngf)) {
|
64
|
emit(context, rootId, dngf.toByteArray());
|
65
|
} else {
|
66
|
//context.getCounter(RelDescriptor.asString(dngf.getRel().getRelType()), "rel marked deleted").increment(1);
|
67
|
}
|
68
|
});
|
69
|
|
70
|
} else {
|
71
|
//context.getCounter(dedupConf.getWf().getEntityType(), "not in duplicate group").increment(1);
|
72
|
|
73
|
final List<String> mergesRels = getTargetIds(value, "merges");
|
74
|
if (mergesRels != null && !mergesRels.isEmpty()) {
|
75
|
final byte[] body = value.getValue(cfMetadataByte(), type.toString().getBytes());
|
76
|
if (body != null) {
|
77
|
context.getCounter(type.toString(), "root").increment(1);
|
78
|
emit(context, new Text(rowkey.copyBytes()), body);
|
79
|
}
|
80
|
}
|
81
|
}
|
82
|
|
83
|
}
|
84
|
|
85
|
private void emit(final Context context, final Text rootId, final byte[] value) {
|
86
|
ibw.set(value);
|
87
|
try {
|
88
|
context.write(rootId, ibw);
|
89
|
} catch (Exception e) {
|
90
|
throw new RuntimeException(e);
|
91
|
}
|
92
|
}
|
93
|
|
94
|
private boolean isRelMarkedDeleted(final DNGF dngf) {
|
95
|
return dngf.getKind().equals(Kind.relation) && dngf.getDataInfo().getDeletedbyinference();
|
96
|
}
|
97
|
|
98
|
}
|