Revision 48800
Added by Claudio Atzori over 6 years ago
| HadoopUtils.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.hadoop.utils; |
| 2 | 2 |
|
| 3 |
import java.util.Map.Entry; |
|
| 4 |
import java.util.Set; |
|
| 5 |
|
|
| 6 |
import org.apache.hadoop.hbase.HColumnDescriptor; |
|
| 7 |
import org.apache.hadoop.hbase.HTableDescriptor; |
|
| 8 |
|
|
| 9 |
import com.google.common.base.Function; |
|
| 10 |
import com.google.common.base.Predicate; |
|
| 11 |
|
|
| 12 | 3 |
import eu.dnetlib.data.hadoop.HadoopJob; |
| 13 |
import eu.dnetlib.data.hadoop.config.ClusterName; |
|
| 14 | 4 |
import eu.dnetlib.data.hadoop.rmi.HadoopJobDescriptor; |
| 15 | 5 |
import eu.dnetlib.data.hadoop.rmi.HadoopServiceException; |
| 6 |
import org.apache.commons.logging.Log; |
|
| 7 |
import org.apache.commons.logging.LogFactory; |
|
| 16 | 8 |
|
| 17 | 9 |
public class HadoopUtils {
|
| 18 | 10 |
|
| 19 |
public static Function<HTableDescriptor, String> tableName() {
|
|
| 20 |
return new Function<HTableDescriptor, String>() {
|
|
| 11 |
private static final Log log = LogFactory.getLog(HadoopUtils.class); |
|
| 21 | 12 |
|
| 22 |
@Override |
|
| 23 |
public String apply(final HTableDescriptor d) {
|
|
| 24 |
return d.getNameAsString(); |
|
| 13 |
public static java.util.function.Function<HadoopJob, HadoopJobDescriptor> asDescriptor() {
|
|
| 14 |
return d -> {
|
|
| 15 |
try {
|
|
| 16 |
return d.asDescriptor(); |
|
| 17 |
} catch (HadoopServiceException e) {
|
|
| 18 |
log.warn(e); |
|
| 19 |
return null; |
|
| 25 | 20 |
} |
| 26 | 21 |
}; |
| 27 | 22 |
} |
| 28 | 23 |
|
| 29 |
public static Function<HColumnDescriptor, String> columnName() {
|
|
| 30 |
return new Function<HColumnDescriptor, String>() {
|
|
| 31 |
|
|
| 32 |
@Override |
|
| 33 |
public String apply(final HColumnDescriptor d) {
|
|
| 34 |
return d.getNameAsString(); |
|
| 35 |
} |
|
| 36 |
}; |
|
| 37 |
} |
|
| 38 |
|
|
| 39 |
public static Predicate<String> columnPredicate(final Set<String> cols) {
|
|
| 40 |
return new HadoopUtils().getSetPredicate(cols); |
|
| 41 |
} |
|
| 42 |
|
|
| 43 |
public SetPredicate getSetPredicate(final Set<String> set) {
|
|
| 44 |
return new SetPredicate(set); |
|
| 45 |
} |
|
| 46 |
|
|
| 47 |
class SetPredicate implements Predicate<String> {
|
|
| 48 |
|
|
| 49 |
private final Set<String> set; |
|
| 50 |
|
|
| 51 |
public SetPredicate(final Set<String> set) {
|
|
| 52 |
this.set = set; |
|
| 53 |
} |
|
| 54 |
|
|
| 55 |
@Override |
|
| 56 |
public boolean apply(final String s) {
|
|
| 57 |
return !set.contains(s); |
|
| 58 |
} |
|
| 59 |
} |
|
| 60 |
|
|
| 61 |
public static Function<Entry<String, HadoopJob>, HadoopJobDescriptor> hadoopJobDescriptor() {
|
|
| 62 |
return new Function<Entry<String, HadoopJob>, HadoopJobDescriptor>() {
|
|
| 63 |
|
|
| 64 |
@Override |
|
| 65 |
public HadoopJobDescriptor apply(final Entry<String, HadoopJob> e) {
|
|
| 66 |
try {
|
|
| 67 |
return e.getValue().asDescriptor(); |
|
| 68 |
} catch (HadoopServiceException e1) {
|
|
| 69 |
return null; |
|
| 70 |
} |
|
| 71 |
} |
|
| 72 |
}; |
|
| 73 |
} |
|
| 74 |
|
|
| 75 |
public static Predicate<HadoopJob> filterByCluster(final ClusterName clusterName) {
|
|
| 76 |
return new Predicate<HadoopJob>() {
|
|
| 77 |
|
|
| 78 |
@Override |
|
| 79 |
public boolean apply(final HadoopJob job) {
|
|
| 80 |
return job.getClusterName().equals(clusterName); |
|
| 81 |
} |
|
| 82 |
|
|
| 83 |
}; |
|
| 84 |
} |
|
| 85 |
|
|
| 86 | 24 |
} |
Also available in: Unified diff
more logging in the oozie job monitor procedure that tries to fetch the job report. More java8 refactorings