Revision 57379
Added by Claudio Atzori over 4 years ago
| EnrichmentMapper.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.mapreduce.hbase.broker.enrich; |
| 2 | 2 |
|
| 3 | 3 |
import java.io.IOException; |
| 4 |
import java.util.List; |
|
| 4 | 5 |
import java.util.Map; |
| 6 |
import java.util.stream.Collectors; |
|
| 5 | 7 |
|
| 6 | 8 |
import com.google.common.collect.Iterables; |
| 7 | 9 |
import eu.dnetlib.data.mapreduce.util.DedupUtils; |
| ... | ... | |
| 54 | 56 |
private byte[] getEmitKey(final Context context, final ImmutableBytesWritable key, final Map<byte[], byte[]> mergedIn) {
|
| 55 | 57 |
if (MapUtils.isNotEmpty(mergedIn)) {
|
| 56 | 58 |
context.getCounter(Type.result.name(), RelName.isMergedIn.name()).increment(1); |
| 57 |
return Iterables.getOnlyElement(mergedIn.keySet()); |
|
| 59 |
|
|
| 60 |
//TODO a duplicate should be merged into only one representative object, but we noticed some cases where |
|
| 61 |
// a duplicate is merged in two different representative objects |
|
| 62 |
return mergedIn.keySet().iterator().next(); |
|
| 58 | 63 |
} else {
|
| 59 | 64 |
return key.copyBytes(); |
| 60 | 65 |
} |
Also available in: Unified diff
pick the first mergedIn identifier