| 1 |
39562
|
michele.ar
|
package eu.dnetlib.data.mapreduce.hbase.broker;
|
| 2 |
|
|
|
| 3 |
|
|
import eu.dnetlib.data.proto.DatasourceProtos;
|
| 4 |
39623
|
michele.ar
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
| 5 |
39562
|
michele.ar
|
import eu.dnetlib.data.proto.PersonProtos;
|
| 6 |
43275
|
claudio.at
|
import org.apache.commons.lang3.StringUtils;
|
| 7 |
39562
|
michele.ar
|
import org.apache.hadoop.hbase.util.Bytes;
|
| 8 |
|
|
import org.apache.hadoop.io.Text;
|
| 9 |
|
|
|
| 10 |
|
|
/**
|
| 11 |
|
|
* Created by michele on 12/10/15.
|
| 12 |
|
|
*/
|
| 13 |
|
|
public class CalculatePersonDistributionUtils {
|
| 14 |
|
|
|
| 15 |
|
|
public static final String SEPARATOR = "||@@||";
|
| 16 |
|
|
private static final String PERSON_PREFIX = "PERSON" + SEPARATOR;
|
| 17 |
|
|
private static final String DS_TYPE_PREFIX = "DSTYPE" + SEPARATOR;
|
| 18 |
|
|
|
| 19 |
39567
|
michele.ar
|
public static String createPersonValue(final PersonProtos.Person p) {
|
| 20 |
39623
|
michele.ar
|
try {
|
| 21 |
|
|
final String fn = p.getMetadata().getFirstname().getValue();
|
| 22 |
|
|
String sn = "";
|
| 23 |
|
|
for (FieldTypeProtos.StringField sf : p.getMetadata().getSecondnamesList()) {
|
| 24 |
|
|
sn += sf.getValue();
|
| 25 |
|
|
}
|
| 26 |
|
|
if (StringUtils.isNotBlank(fn) && StringUtils.isNotBlank(sn)) {
|
| 27 |
|
|
return PERSON_PREFIX + sn + "_" + fn.substring(0, 1);
|
| 28 |
|
|
}
|
| 29 |
|
|
return null;
|
| 30 |
|
|
} catch (Throwable e) {
|
| 31 |
|
|
return null;
|
| 32 |
|
|
}
|
| 33 |
39562
|
michele.ar
|
}
|
| 34 |
|
|
|
| 35 |
39567
|
michele.ar
|
public static String createDsTypeValue(final DatasourceProtos.Datasource ds) {
|
| 36 |
|
|
return DS_TYPE_PREFIX + ds.getMetadata().getDatasourcetype().getClassid();
|
| 37 |
39562
|
michele.ar
|
}
|
| 38 |
|
|
|
| 39 |
|
|
public static boolean isPerson(byte[] b) {
|
| 40 |
|
|
return Bytes.toString(b).startsWith(PERSON_PREFIX);
|
| 41 |
|
|
}
|
| 42 |
|
|
|
| 43 |
|
|
public static boolean isDsType(byte[] b) {
|
| 44 |
|
|
return Bytes.toString(b).startsWith(DS_TYPE_PREFIX);
|
| 45 |
|
|
}
|
| 46 |
|
|
|
| 47 |
|
|
public static String getPersonName(final byte[] b) {
|
| 48 |
|
|
return StringUtils.substringAfter(Bytes.toString(b), PERSON_PREFIX);
|
| 49 |
|
|
}
|
| 50 |
|
|
|
| 51 |
|
|
public static String getType(final byte[] b) {
|
| 52 |
|
|
return StringUtils.substringAfter(Bytes.toString(b), DS_TYPE_PREFIX);
|
| 53 |
|
|
}
|
| 54 |
|
|
|
| 55 |
|
|
public static Text prepareHdfsValue(String dsId, String name) {
|
| 56 |
|
|
return new Text(dsId + CalculatePersonDistributionUtils.SEPARATOR + name);
|
| 57 |
|
|
}
|
| 58 |
|
|
|
| 59 |
|
|
public static Text prepareHdfsKey(final String dsId, final int count) {
|
| 60 |
|
|
return new Text(dsId + "_" + count);
|
| 61 |
|
|
}
|
| 62 |
|
|
|
| 63 |
39567
|
michele.ar
|
public static Text getDsIdFromHdfsValue(Text text) {
|
| 64 |
|
|
return new Text(StringUtils.substringBefore(Bytes.toString(text.copyBytes()), SEPARATOR));
|
| 65 |
39562
|
michele.ar
|
}
|
| 66 |
|
|
|
| 67 |
|
|
public static Text getNameFromHdfsValue(Text text) {
|
| 68 |
|
|
return new Text(StringUtils.substringAfter(Bytes.toString(text.copyBytes()), SEPARATOR));
|
| 69 |
|
|
}
|
| 70 |
|
|
|
| 71 |
|
|
}
|