|
1
|
//package eu.dnetlib.data.mapreduce.hbase.dedup;
|
|
2
|
//
|
|
3
|
//import java.io.IOException;
|
|
4
|
//import java.nio.ByteBuffer;
|
|
5
|
//import java.util.Map;
|
|
6
|
//
|
|
7
|
//import org.apache.hadoop.hbase.client.Result;
|
|
8
|
//import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
|
|
9
|
//import org.apache.hadoop.hbase.mapreduce.TableMapper;
|
|
10
|
//import org.apache.hadoop.hbase.util.Bytes;
|
|
11
|
//
|
|
12
|
//import com.google.protobuf.InvalidProtocolBufferException;
|
|
13
|
//import com.googlecode.protobuf.format.JsonFormat;
|
|
14
|
//
|
|
15
|
//import eu.dnetlib.data.mapreduce.JobParams;
|
|
16
|
//import eu.dnetlib.data.mapreduce.util.DedupUtils;
|
|
17
|
//import eu.dnetlib.data.mapreduce.util.DNGFDecoder;
|
|
18
|
//import eu.dnetlib.data.mapreduce.util.DNGFRowKeyDecoder;
|
|
19
|
//import eu.dnetlib.data.proto.DNGFProtos.DNGF;
|
|
20
|
//import eu.dnetlib.data.proto.PersonProtos.Person;
|
|
21
|
//import eu.dnetlib.data.proto.TypeProtos.Type;
|
|
22
|
//import eu.dnetlib.pace.config.DedupConfig;
|
|
23
|
//import eu.dnetlib.pace.model.gt.GTAuthor;
|
|
24
|
//
|
|
25
|
//public class DedupFindRootsPersonMapper extends TableMapper<ImmutableBytesWritable, ImmutableBytesWritable> {
|
|
26
|
//
|
|
27
|
// private DedupConfig dedupConf;
|
|
28
|
//
|
|
29
|
// private ImmutableBytesWritable outKey;
|
|
30
|
//
|
|
31
|
// private ImmutableBytesWritable outValue;
|
|
32
|
//
|
|
33
|
// @Override
|
|
34
|
// protected void setup(final Context context) throws IOException, InterruptedException {
|
|
35
|
// dedupConf = DedupConfig.load(context.getConfiguration().get(JobParams.DEDUP_CONF));
|
|
36
|
// System.out.println("dedup findRoots mapper\nwf conf: " + dedupConf.toString());
|
|
37
|
//
|
|
38
|
// outKey = new ImmutableBytesWritable();
|
|
39
|
// outValue = new ImmutableBytesWritable();
|
|
40
|
// }
|
|
41
|
//
|
|
42
|
// @Override
|
|
43
|
// protected void map(final ImmutableBytesWritable rowkey, final Result value, final Context context) throws IOException, InterruptedException {
|
|
44
|
// // System.out.println("Find root mapping: " + new String(rowkey.copyBytes()));
|
|
45
|
//
|
|
46
|
// final DNGFRowKeyDecoder rkd = DNGFRowKeyDecoder.decode(rowkey.copyBytes());
|
|
47
|
//
|
|
48
|
// if (!Type.person.equals(rkd.getType())) {
|
|
49
|
// context.getCounter(rkd.getType().toString(), "skipped").increment(1);
|
|
50
|
// }
|
|
51
|
//
|
|
52
|
// final Type type = Type.valueOf(dedupConf.getWf().getEntityType());
|
|
53
|
// final Map<byte[], byte[]> similarRels = value.getFamilyMap(DedupUtils.getSimilarityCFBytes(type));
|
|
54
|
//
|
|
55
|
// if ((similarRels != null) && !similarRels.isEmpty()) {
|
|
56
|
// final ByteBuffer min = findMin(rowkey.copyBytes(), similarRels.keySet());
|
|
57
|
//
|
|
58
|
// final byte[] groupingKey = DedupUtils.newIdBytes(min, dedupConf.getWf().getDedupRun());
|
|
59
|
//
|
|
60
|
// final GTAuthor gta = asGTA(rowkey, value.getValue(Bytes.toBytes(dedupConf.getWf().getEntityType()), DedupUtils.BODY_B));
|
|
61
|
//
|
|
62
|
// emitBody(context, groupingKey, gta);
|
|
63
|
// } else {
|
|
64
|
// context.getCounter(dedupConf.getWf().getEntityType(), "row not in similarity mesh").increment(1);
|
|
65
|
// }
|
|
66
|
// }
|
|
67
|
//
|
|
68
|
// private GTAuthor asGTA(final ImmutableBytesWritable rowkey, final byte[] input) {
|
|
69
|
//
|
|
70
|
// final DNGFDecoder decoder = DNGFDecoder.decode(input);
|
|
71
|
// final DNGF oaf = decoder.getDNGF();
|
|
72
|
//
|
|
73
|
// final Person person = oaf.getEntity().getPerson();
|
|
74
|
//
|
|
75
|
// final GTAuthor gta = GTAuthor.fromDNGFJson(new JsonFormat().printToString(person));
|
|
76
|
// final String id = new String(rowkey.copyBytes());
|
|
77
|
// gta.setId(id);
|
|
78
|
// gta.getAuthor().setId(id);
|
|
79
|
// return gta;
|
|
80
|
// }
|
|
81
|
//
|
|
82
|
// private ByteBuffer findMin(final byte[] key, final Iterable<byte[]> keys) {
|
|
83
|
// ByteBuffer bb = ByteBuffer.wrap(key);
|
|
84
|
// for (final byte[] q : keys) {
|
|
85
|
// final ByteBuffer iq = ByteBuffer.wrap(q);
|
|
86
|
// if (bb.compareTo(iq) > 0) {
|
|
87
|
// bb = iq;
|
|
88
|
// }
|
|
89
|
// }
|
|
90
|
// return bb;
|
|
91
|
// }
|
|
92
|
//
|
|
93
|
// private void emitBody(final Context context, final byte[] row, final GTAuthor gta) throws IOException, InterruptedException {
|
|
94
|
//
|
|
95
|
// outKey.set(row);
|
|
96
|
// outValue.set(toDNGFByteArray(gta));
|
|
97
|
//
|
|
98
|
// context.write(outKey, outValue);
|
|
99
|
// context.getCounter(dedupConf.getWf().getEntityType(), "in").increment(1);
|
|
100
|
// }
|
|
101
|
//
|
|
102
|
// public byte[] toDNGFByteArray(final GTAuthor gta) {
|
|
103
|
// // final DNGF oaf = new GTAuthorMapper().map(gta);
|
|
104
|
// // return oaf.toByteArray();
|
|
105
|
// return Bytes.toBytes(gta.toString());
|
|
106
|
// }
|
|
107
|
//
|
|
108
|
//}
|