2020-09-17 11:38:07 +02:00
|
|
|
|
|
|
|
package eu.dnetlib.dhp.broker.oa;
|
|
|
|
|
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
|
|
|
import java.io.IOException;
|
2020-12-07 14:56:06 +01:00
|
|
|
import java.util.Arrays;
|
|
|
|
import java.util.HashSet;
|
2020-09-17 11:38:07 +02:00
|
|
|
import java.util.Optional;
|
2020-12-07 14:56:06 +01:00
|
|
|
import java.util.Set;
|
|
|
|
import java.util.stream.Collectors;
|
2020-09-17 11:38:07 +02:00
|
|
|
|
2020-12-07 14:56:06 +01:00
|
|
|
import org.apache.commons.codec.digest.DigestUtils;
|
2020-09-17 11:38:07 +02:00
|
|
|
import org.apache.commons.io.IOUtils;
|
|
|
|
import org.apache.commons.lang3.StringUtils;
|
|
|
|
import org.apache.hadoop.conf.Configuration;
|
|
|
|
import org.apache.hadoop.fs.FileStatus;
|
|
|
|
import org.apache.hadoop.fs.FileSystem;
|
|
|
|
import org.apache.hadoop.fs.Path;
|
|
|
|
import org.apache.spark.SparkConf;
|
2020-12-07 21:53:13 +01:00
|
|
|
import org.apache.spark.api.java.function.FilterFunction;
|
|
|
|
import org.apache.spark.api.java.function.MapFunction;
|
2020-09-17 11:38:07 +02:00
|
|
|
import org.apache.spark.sql.Encoders;
|
|
|
|
import org.apache.spark.sql.SaveMode;
|
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
|
|
|
|
|
|
|
import com.google.gson.Gson;
|
|
|
|
|
|
|
|
import eu.dnetlib.broker.objects.OaBrokerEventPayload;
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
|
|
import eu.dnetlib.dhp.broker.model.Event;
|
2020-09-25 09:02:58 +02:00
|
|
|
import eu.dnetlib.dhp.broker.model.ShortEventMessageWithGroupId;
|
2020-09-17 11:38:07 +02:00
|
|
|
import eu.dnetlib.dhp.broker.oa.util.ClusterUtils;
|
|
|
|
|
|
|
|
public class PartitionEventsByDsIdJob {
|
|
|
|
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(PartitionEventsByDsIdJob.class);
|
2021-01-14 13:20:35 +01:00
|
|
|
private static final String OPENDOAR_NSPREFIX = "opendoar____::";
|
2020-09-17 11:38:07 +02:00
|
|
|
|
|
|
|
public static void main(final String[] args) throws Exception {
|
|
|
|
|
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
|
|
|
|
IOUtils
|
2020-12-10 14:47:51 +01:00
|
|
|
.toString(
|
|
|
|
PartitionEventsByDsIdJob.class
|
|
|
|
.getResourceAsStream("/eu/dnetlib/dhp/broker/oa/od_partitions_params.json")));
|
2020-09-17 11:38:07 +02:00
|
|
|
parser.parseArgument(args);
|
|
|
|
|
|
|
|
final Boolean isSparkSessionManaged = Optional
|
|
|
|
.ofNullable(parser.get("isSparkSessionManaged"))
|
|
|
|
.map(Boolean::valueOf)
|
|
|
|
.orElse(Boolean.TRUE);
|
|
|
|
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
|
|
|
|
|
|
|
final SparkConf conf = new SparkConf();
|
|
|
|
|
2020-12-10 14:47:22 +01:00
|
|
|
final String eventsPath = parser.get("outputDir") + "/events";
|
2020-09-17 11:38:07 +02:00
|
|
|
log.info("eventsPath: {}", eventsPath);
|
|
|
|
|
2020-12-10 14:47:22 +01:00
|
|
|
final String partitionPath = parser.get("outputDir") + "/eventsByOpendoarId";
|
2020-09-17 11:38:07 +02:00
|
|
|
log.info("partitionPath: {}", partitionPath);
|
|
|
|
|
2020-12-07 14:56:06 +01:00
|
|
|
final String opendoarIds = parser.get("opendoarIds");
|
|
|
|
log.info("opendoarIds: {}", opendoarIds);
|
|
|
|
|
|
|
|
final Set<String> validOpendoarIds = new HashSet<>();
|
|
|
|
if (!opendoarIds.trim().equals("-")) {
|
2020-12-07 21:53:13 +01:00
|
|
|
validOpendoarIds
|
2020-12-10 14:47:51 +01:00
|
|
|
.addAll(
|
|
|
|
Arrays
|
|
|
|
.stream(opendoarIds.split(","))
|
|
|
|
.map(String::trim)
|
|
|
|
.filter(StringUtils::isNotBlank)
|
|
|
|
.map(s -> OPENDOAR_NSPREFIX + DigestUtils.md5Hex(s))
|
|
|
|
.collect(Collectors.toSet()));
|
2020-12-07 14:56:06 +01:00
|
|
|
}
|
2020-12-07 21:53:13 +01:00
|
|
|
log.info("validOpendoarIds: {}", validOpendoarIds);
|
2020-12-07 14:56:06 +01:00
|
|
|
|
2020-09-17 11:38:07 +02:00
|
|
|
runWithSparkSession(conf, isSparkSessionManaged, spark -> {
|
|
|
|
|
|
|
|
ClusterUtils
|
|
|
|
.readPath(spark, eventsPath, Event.class)
|
2020-12-07 21:53:13 +01:00
|
|
|
.filter((FilterFunction<Event>) e -> StringUtils.isNotBlank(e.getMap().getTargetDatasourceId()))
|
|
|
|
.filter((FilterFunction<Event>) e -> e.getMap().getTargetDatasourceId().startsWith(OPENDOAR_NSPREFIX))
|
|
|
|
.filter((FilterFunction<Event>) e -> validOpendoarIds.contains(e.getMap().getTargetDatasourceId()))
|
2020-12-10 14:47:51 +01:00
|
|
|
.map(
|
|
|
|
(MapFunction<Event, ShortEventMessageWithGroupId>) e -> messageFromNotification(e),
|
|
|
|
Encoders.bean(ShortEventMessageWithGroupId.class))
|
2020-09-25 09:02:58 +02:00
|
|
|
.coalesce(1)
|
2020-09-17 11:38:07 +02:00
|
|
|
.write()
|
2020-09-25 09:02:58 +02:00
|
|
|
.partitionBy("group")
|
2020-09-17 11:38:07 +02:00
|
|
|
.mode(SaveMode.Overwrite)
|
2020-12-10 11:59:28 +01:00
|
|
|
.option("compression", "gzip")
|
2020-09-17 11:38:07 +02:00
|
|
|
.json(partitionPath);
|
|
|
|
|
|
|
|
});
|
|
|
|
renameSubDirs(partitionPath);
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void renameSubDirs(final String path) throws IOException {
|
|
|
|
final FileSystem fs = FileSystem.get(new Configuration());
|
|
|
|
|
|
|
|
log.info("** Renaming subdirs of " + path);
|
|
|
|
for (final FileStatus fileStatus : fs.listStatus(new Path(path))) {
|
|
|
|
if (fileStatus.isDirectory()) {
|
|
|
|
final Path oldPath = fileStatus.getPath();
|
|
|
|
final String oldName = oldPath.getName();
|
2020-09-25 09:02:58 +02:00
|
|
|
if (oldName.contains("=")) {
|
|
|
|
final Path newPath = new Path(path + "/" + StringUtils.substringAfter(oldName, "="));
|
2020-09-17 11:38:07 +02:00
|
|
|
log.info(" * " + oldPath.getName() + " -> " + newPath.getName());
|
|
|
|
fs.rename(oldPath, newPath);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-09-25 09:02:58 +02:00
|
|
|
private static ShortEventMessageWithGroupId messageFromNotification(final Event e) {
|
2020-09-17 11:38:07 +02:00
|
|
|
final Gson gson = new Gson();
|
|
|
|
|
|
|
|
final OaBrokerEventPayload payload = gson.fromJson(e.getPayload(), OaBrokerEventPayload.class);
|
|
|
|
|
2020-09-25 09:02:58 +02:00
|
|
|
final ShortEventMessageWithGroupId res = new ShortEventMessageWithGroupId();
|
2020-09-17 11:38:07 +02:00
|
|
|
|
2021-01-14 13:20:35 +01:00
|
|
|
res.setEventId(e.getEventId());
|
2020-09-17 11:38:07 +02:00
|
|
|
res.setOriginalId(payload.getResult().getOriginalId());
|
|
|
|
res.setTitle(payload.getResult().getTitles().stream().filter(StringUtils::isNotBlank).findFirst().orElse(null));
|
|
|
|
res.setTopic(e.getTopic());
|
|
|
|
res.setTrust(payload.getTrust());
|
|
|
|
res.generateMessageFromObject(payload.getHighlight());
|
2020-09-25 09:02:58 +02:00
|
|
|
res.setGroup(StringUtils.substringAfter(e.getMap().getTargetDatasourceId(), OPENDOAR_NSPREFIX));
|
2020-09-17 11:38:07 +02:00
|
|
|
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|