1 |
53588
|
sandro.lab
|
package eu.dnetlib.data.mapreduce.hbase.dataexport;
|
2 |
|
|
|
3 |
53604
|
claudio.at
|
import com.google.common.base.Splitter;
|
4 |
53628
|
claudio.at
|
import com.google.common.collect.Lists;
|
5 |
53604
|
claudio.at
|
import com.google.common.collect.Sets;
|
6 |
53588
|
sandro.lab
|
import com.googlecode.protobuf.format.JsonFormat;
|
7 |
|
|
import eu.dnetlib.data.mapreduce.util.DedupUtils;
|
8 |
|
|
import eu.dnetlib.data.proto.OafProtos;
|
9 |
|
|
import eu.dnetlib.data.proto.TypeProtos;
|
10 |
|
|
import org.apache.hadoop.hbase.client.Result;
|
11 |
|
|
import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
|
12 |
|
|
import org.apache.hadoop.hbase.mapreduce.TableMapper;
|
13 |
|
|
import org.apache.hadoop.io.Text;
|
14 |
53604
|
claudio.at
|
import org.apache.hadoop.mapreduce.Mapper;
|
15 |
53588
|
sandro.lab
|
|
16 |
|
|
import java.io.IOException;
|
17 |
|
|
import java.util.Map;
|
18 |
53604
|
claudio.at
|
import java.util.Set;
|
19 |
53588
|
sandro.lab
|
|
20 |
53604
|
claudio.at
|
public class ConfigurableExportMapper extends TableMapper<Text, Text> {
|
21 |
53588
|
sandro.lab
|
|
22 |
53604
|
claudio.at
|
private static final String CF_DEFAULT = "result";
|
23 |
|
|
private static final String QUALIFIER_DEFAULT = "body";
|
24 |
|
|
|
25 |
53588
|
sandro.lab
|
private Text keyOut;
|
26 |
|
|
private Text valueOut;
|
27 |
|
|
|
28 |
53604
|
claudio.at
|
private Set<String> columnFamilies = Sets.newHashSet();
|
29 |
|
|
private Set<String> qualifiers = Sets.newHashSet();
|
30 |
|
|
|
31 |
|
|
enum ExportMode {
|
32 |
|
|
ALL,
|
33 |
|
|
DUPS_ONLY
|
34 |
|
|
}
|
35 |
|
|
|
36 |
|
|
private ExportMode exportMode = ExportMode.ALL;
|
37 |
|
|
|
38 |
53588
|
sandro.lab
|
@Override
|
39 |
|
|
protected void setup(Context context) throws IOException, InterruptedException {
|
40 |
|
|
keyOut = new Text("");
|
41 |
|
|
valueOut = new Text();
|
42 |
53604
|
claudio.at
|
|
43 |
|
|
|
44 |
53628
|
claudio.at
|
columnFamilies.addAll(Lists.newArrayList(Splitter.on(",").split(context.getConfiguration().get("column.families", CF_DEFAULT))));
|
45 |
|
|
qualifiers.addAll(Lists.newArrayList(Splitter.on(",").split(context.getConfiguration().get("qualifiers", QUALIFIER_DEFAULT))));
|
46 |
53604
|
claudio.at
|
exportMode = ExportMode.valueOf(context.getConfiguration().get("export.mode", ExportMode.ALL.toString()));
|
47 |
|
|
|
48 |
53588
|
sandro.lab
|
}
|
49 |
|
|
|
50 |
|
|
@Override
|
51 |
|
|
protected void map(ImmutableBytesWritable key, Result value, Context context) throws IOException, InterruptedException {
|
52 |
53604
|
claudio.at
|
for(String cf : columnFamilies) {
|
53 |
|
|
switch (exportMode) {
|
54 |
|
|
case ALL:
|
55 |
|
|
export(value, context, cf);
|
56 |
|
|
break;
|
57 |
|
|
case DUPS_ONLY:
|
58 |
|
|
final Map<byte[], byte[]> mergedInMap = value.getFamilyMap(DedupUtils.getDedupCF_mergedInBytes(cf));
|
59 |
|
|
if (mergedInMap != null && !mergedInMap.isEmpty()) {
|
60 |
|
|
export(value, context, cf);
|
61 |
|
|
}
|
62 |
53588
|
sandro.lab
|
|
63 |
53604
|
claudio.at
|
break;
|
64 |
|
|
}
|
65 |
|
|
}
|
66 |
|
|
}
|
67 |
53588
|
sandro.lab
|
|
68 |
53604
|
claudio.at
|
private void export(Result value, Context context, String cf) throws IOException, InterruptedException {
|
69 |
|
|
for(String q : qualifiers) {
|
70 |
53588
|
sandro.lab
|
|
71 |
53604
|
claudio.at
|
final byte[] data = value.getValue(cf.getBytes(), q.getBytes());
|
72 |
|
|
|
73 |
|
|
if (data != null) {
|
74 |
|
|
OafProtos.Oaf oaf = OafProtos.Oaf.parseFrom(data);
|
75 |
53588
|
sandro.lab
|
valueOut.set(JsonFormat.printToString(oaf));
|
76 |
|
|
context.write(keyOut, valueOut);
|
77 |
53604
|
claudio.at
|
|
78 |
|
|
context.getCounter(cf, q).increment(1);
|
79 |
53588
|
sandro.lab
|
}
|
80 |
|
|
}
|
81 |
53604
|
claudio.at
|
}
|
82 |
53588
|
sandro.lab
|
|
83 |
53604
|
claudio.at
|
|
84 |
53588
|
sandro.lab
|
}
|