2020-06-22 11:45:14 +02:00
|
|
|
|
|
|
|
package eu.dnetlib.dhp.broker.oa;
|
|
|
|
|
|
|
|
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
|
|
|
|
|
2020-06-29 16:33:32 +02:00
|
|
|
import java.util.Map;
|
2020-06-22 11:45:14 +02:00
|
|
|
import java.util.Optional;
|
2020-06-29 16:33:32 +02:00
|
|
|
import java.util.stream.Collectors;
|
2020-06-22 11:45:14 +02:00
|
|
|
|
|
|
|
import org.apache.commons.io.IOUtils;
|
|
|
|
import org.apache.spark.SparkConf;
|
2020-06-29 16:33:32 +02:00
|
|
|
import org.apache.spark.SparkContext;
|
2020-06-22 11:45:14 +02:00
|
|
|
import org.apache.spark.sql.Dataset;
|
|
|
|
import org.apache.spark.sql.Encoders;
|
2020-06-29 16:33:32 +02:00
|
|
|
import org.apache.spark.util.LongAccumulator;
|
2020-06-22 11:45:14 +02:00
|
|
|
import org.slf4j.Logger;
|
|
|
|
import org.slf4j.LoggerFactory;
|
|
|
|
|
|
|
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
|
|
|
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser;
|
|
|
|
import eu.dnetlib.dhp.broker.model.Event;
|
2020-06-29 16:33:32 +02:00
|
|
|
import eu.dnetlib.dhp.broker.oa.matchers.UpdateMatcher;
|
2020-06-22 11:45:14 +02:00
|
|
|
import eu.dnetlib.dhp.broker.oa.util.ClusterUtils;
|
|
|
|
import eu.dnetlib.dhp.broker.oa.util.EventFinder;
|
|
|
|
import eu.dnetlib.dhp.broker.oa.util.EventGroup;
|
|
|
|
import eu.dnetlib.dhp.broker.oa.util.aggregators.simple.ResultGroup;
|
|
|
|
import eu.dnetlib.dhp.utils.ISLookupClientFactory;
|
|
|
|
import eu.dnetlib.enabling.is.lookup.rmi.ISLookUpService;
|
|
|
|
import eu.dnetlib.pace.config.DedupConfig;
|
|
|
|
|
|
|
|
public class GenerateEventsJob {
|
|
|
|
|
|
|
|
private static final Logger log = LoggerFactory.getLogger(GenerateEventsJob.class);
|
|
|
|
|
|
|
|
public static void main(final String[] args) throws Exception {
|
|
|
|
final ArgumentApplicationParser parser = new ArgumentApplicationParser(
|
|
|
|
IOUtils
|
|
|
|
.toString(
|
|
|
|
GenerateEventsJob.class
|
|
|
|
.getResourceAsStream("/eu/dnetlib/dhp/broker/oa/generate_events.json")));
|
|
|
|
parser.parseArgument(args);
|
|
|
|
|
|
|
|
final Boolean isSparkSessionManaged = Optional
|
|
|
|
.ofNullable(parser.get("isSparkSessionManaged"))
|
|
|
|
.map(Boolean::valueOf)
|
|
|
|
.orElse(Boolean.TRUE);
|
|
|
|
log.info("isSparkSessionManaged: {}", isSparkSessionManaged);
|
|
|
|
|
|
|
|
final String workingPath = parser.get("workingPath");
|
|
|
|
log.info("workingPath: {}", workingPath);
|
|
|
|
|
|
|
|
final String isLookupUrl = parser.get("isLookupUrl");
|
|
|
|
log.info("isLookupUrl: {}", isLookupUrl);
|
|
|
|
|
|
|
|
final String dedupConfigProfileId = parser.get("dedupConfProfile");
|
|
|
|
log.info("dedupConfigProfileId: {}", dedupConfigProfileId);
|
|
|
|
|
2020-06-24 09:24:45 +02:00
|
|
|
final String eventsPath = workingPath + "/events";
|
2020-06-22 11:45:14 +02:00
|
|
|
log.info("eventsPath: {}", eventsPath);
|
|
|
|
|
|
|
|
final SparkConf conf = new SparkConf();
|
|
|
|
|
|
|
|
// TODO UNCOMMENT
|
|
|
|
// final DedupConfig dedupConfig = loadDedupConfig(isLookupUrl, dedupConfigProfileId);
|
|
|
|
final DedupConfig dedupConfig = null;
|
|
|
|
|
|
|
|
runWithSparkSession(conf, isSparkSessionManaged, spark -> {
|
|
|
|
|
|
|
|
ClusterUtils.removeDir(spark, eventsPath);
|
|
|
|
|
2020-06-29 16:33:32 +02:00
|
|
|
final Map<String, LongAccumulator> accumulators = prepareAccumulators(spark.sparkContext());
|
|
|
|
|
2020-06-30 16:17:09 +02:00
|
|
|
final LongAccumulator total = spark.sparkContext().longAccumulator("total_events");
|
|
|
|
|
2020-06-22 11:45:14 +02:00
|
|
|
final Dataset<ResultGroup> groups = ClusterUtils
|
2020-06-24 08:57:13 +02:00
|
|
|
.readPath(spark, workingPath + "/duplicates", ResultGroup.class);
|
2020-06-22 11:45:14 +02:00
|
|
|
|
2020-06-30 16:17:09 +02:00
|
|
|
final Dataset<Event> dataset = groups
|
|
|
|
.map(g -> EventFinder.generateEvents(g, dedupConfig, accumulators), Encoders.bean(EventGroup.class))
|
|
|
|
.flatMap(g -> g.getData().iterator(), Encoders.bean(Event.class))
|
|
|
|
.map(e -> ClusterUtils.incrementAccumulator(e, total), Encoders.bean(Event.class));
|
2020-06-22 11:45:14 +02:00
|
|
|
|
2020-06-30 16:17:09 +02:00
|
|
|
ClusterUtils.save(dataset, eventsPath, Event.class, total);
|
2020-06-22 11:45:14 +02:00
|
|
|
|
|
|
|
});
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2020-06-29 16:33:32 +02:00
|
|
|
public static Map<String, LongAccumulator> prepareAccumulators(final SparkContext sc) {
|
|
|
|
|
|
|
|
return EventFinder
|
|
|
|
.getMatchers()
|
|
|
|
.stream()
|
|
|
|
.map(UpdateMatcher::accumulatorName)
|
|
|
|
.distinct()
|
|
|
|
.collect(Collectors.toMap(s -> s, s -> sc.longAccumulator(s)));
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2020-06-22 11:45:14 +02:00
|
|
|
private static DedupConfig loadDedupConfig(final String isLookupUrl, final String profId) throws Exception {
|
|
|
|
|
|
|
|
final ISLookUpService isLookUpService = ISLookupClientFactory.getLookUpService(isLookupUrl);
|
|
|
|
|
|
|
|
final String conf = isLookUpService
|
|
|
|
.getResourceProfileByQuery(
|
|
|
|
String
|
|
|
|
.format(
|
|
|
|
"for $x in /RESOURCE_PROFILE[.//RESOURCE_IDENTIFIER/@value = '%s'] return $x//DEDUPLICATION/text()",
|
|
|
|
profId));
|
|
|
|
|
|
|
|
final DedupConfig dedupConfig = new ObjectMapper().readValue(conf, DedupConfig.class);
|
|
|
|
dedupConfig.getPace().initModel();
|
|
|
|
dedupConfig.getPace().initTranslationMap();
|
|
|
|
// dedupConfig.getWf().setConfigurationId("???");
|
|
|
|
|
|
|
|
return dedupConfig;
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|