diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/CheckDuplictedIdsJob.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/CheckDuplictedIdsJob.java new file mode 100644 index 000000000..5ca865e8f --- /dev/null +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/CheckDuplictedIdsJob.java @@ -0,0 +1,112 @@ + +package eu.dnetlib.dhp.broker.oa; + +import org.apache.commons.io.IOUtils; +import org.apache.commons.lang.StringUtils; +import org.apache.spark.SparkConf; +import org.apache.spark.sql.Encoder; +import org.apache.spark.sql.Encoders; +import org.apache.spark.sql.SaveMode; +import org.apache.spark.sql.SparkSession; +import org.apache.spark.sql.TypedColumn; +import org.apache.spark.sql.expressions.Aggregator; +import org.apache.spark.util.LongAccumulator; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +import com.fasterxml.jackson.core.JsonProcessingException; +import com.fasterxml.jackson.databind.ObjectMapper; + +import eu.dnetlib.dhp.application.ArgumentApplicationParser; +import eu.dnetlib.dhp.broker.model.Event; +import eu.dnetlib.dhp.broker.oa.util.ClusterUtils; +import scala.Tuple2; + +public class CheckDuplictedIdsJob { + + private static final Logger log = LoggerFactory.getLogger(CheckDuplictedIdsJob.class); + + public static void main(final String[] args) throws Exception { + + final ArgumentApplicationParser parser = new ArgumentApplicationParser( + IOUtils + .toString( + CheckDuplictedIdsJob.class + .getResourceAsStream("/eu/dnetlib/dhp/broker/oa/common_params.json"))); + parser.parseArgument(args); + + final SparkConf conf = new SparkConf(); + + final String eventsPath = parser.get("workingPath") + "/events"; + log.info("eventsPath: {}", eventsPath); + + final String countPath = parser.get("workingPath") + "/counts"; + log.info("countPath: {}", countPath); + + final SparkSession spark = SparkSession.builder().config(conf).getOrCreate(); + + final LongAccumulator total = spark.sparkContext().longAccumulator("invaild_event_id"); + + final TypedColumn, Tuple2> agg = new CountAggregator().toColumn(); + + ClusterUtils + .readPath(spark, eventsPath, Event.class) + .map(e -> new Tuple2<>(e.getEventId(), 1l), Encoders.tuple(Encoders.STRING(), Encoders.LONG())) + .groupByKey(t -> t._1, Encoders.STRING()) + .agg(agg) + .map(t -> t._2, Encoders.tuple(Encoders.STRING(), Encoders.LONG())) + .filter(t -> t._2 > 1) + .map(o -> ClusterUtils.incrementAccumulator(o, total), Encoders.tuple(Encoders.STRING(), Encoders.LONG())) + .write() + .mode(SaveMode.Overwrite) + .json(countPath); + ; + + } + + private static String eventAsJsonString(final Event f) throws JsonProcessingException { + return new ObjectMapper().writeValueAsString(f); + } + +} + +class CountAggregator extends Aggregator, Tuple2, Tuple2> { + + /** + * + */ + private static final long serialVersionUID = 1395935985734672538L; + + @Override + public Encoder> bufferEncoder() { + return Encoders.tuple(Encoders.STRING(), Encoders.LONG()); + } + + @Override + public Tuple2 finish(final Tuple2 arg0) { + return arg0; + } + + @Override + public Tuple2 merge(final Tuple2 arg0, final Tuple2 arg1) { + final String s = StringUtils.defaultIfBlank(arg0._1, arg1._1); + return new Tuple2<>(s, arg0._2 + arg1._2); + } + + @Override + public Encoder> outputEncoder() { + return Encoders.tuple(Encoders.STRING(), Encoders.LONG()); + } + + @Override + public Tuple2 reduce(final Tuple2 arg0, final Tuple2 arg1) { + final String s = StringUtils.defaultIfBlank(arg0._1, arg1._1); + return new Tuple2<>(s, arg0._2 + arg1._2); + } + + @Override + public Tuple2 zero() { + return new Tuple2<>(null, 0l); + } + +} diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/GenerateEventsJob.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/GenerateEventsJob.java index 5d3121aed..a2d92e149 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/GenerateEventsJob.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/GenerateEventsJob.java @@ -3,8 +3,6 @@ package eu.dnetlib.dhp.broker.oa; import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession; -import java.util.Arrays; -import java.util.HashSet; import java.util.Map; import java.util.Optional; import java.util.Set; @@ -63,13 +61,13 @@ public class GenerateEventsJob { final String eventsPath = workingPath + "/events"; log.info("eventsPath: {}", eventsPath); - final Set dsIdWhitelist = parseParamAsList(parser, "datasourceIdWhitelist"); + final Set dsIdWhitelist = ClusterUtils.parseParamAsList(parser, "datasourceIdWhitelist"); log.info("datasourceIdWhitelist: {}", StringUtils.join(dsIdWhitelist, ",")); - final Set dsTypeWhitelist = parseParamAsList(parser, "datasourceTypeWhitelist"); + final Set dsTypeWhitelist = ClusterUtils.parseParamAsList(parser, "datasourceTypeWhitelist"); log.info("datasourceTypeWhitelist: {}", StringUtils.join(dsTypeWhitelist, ",")); - final Set dsIdBlacklist = parseParamAsList(parser, "datasourceIdBlacklist"); + final Set dsIdBlacklist = ClusterUtils.parseParamAsList(parser, "datasourceIdBlacklist"); log.info("datasourceIdBlacklist: {}", StringUtils.join(dsIdBlacklist, ",")); final SparkConf conf = new SparkConf(); @@ -103,22 +101,6 @@ public class GenerateEventsJob { } - private static Set parseParamAsList(final ArgumentApplicationParser parser, final String key) { - final String s = parser.get(key).trim(); - - final Set res = new HashSet<>(); - - if (s.length() > 1) { // A value of a single char (for example: '-') indicates an empty list - Arrays - .stream(s.split(",")) - .map(String::trim) - .filter(StringUtils::isNotBlank) - .forEach(res::add); - } - - return res; - } - public static Map prepareAccumulators(final SparkContext sc) { return EventFinder diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/IndexOnESJob.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/IndexOnESJob.java index 9124d18e3..806147bdd 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/IndexOnESJob.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/IndexOnESJob.java @@ -48,7 +48,7 @@ public class IndexOnESJob { final JavaRDD inputRdd = ClusterUtils .readPath(spark, eventsPath, Event.class) - .limit(10000) // TODO REMOVE + // .limit(10000) // TODO REMOVE .map(IndexOnESJob::eventAsJsonString, Encoders.STRING()) .javaRDD(); diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/relatedProjects/EnrichMoreProject.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/relatedProjects/EnrichMoreProject.java index c38fd81a6..85086a6df 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/relatedProjects/EnrichMoreProject.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/matchers/relatedProjects/EnrichMoreProject.java @@ -25,7 +25,9 @@ public class EnrichMoreProject extends UpdateMatcher { protected List findDifferences(final OaBrokerMainEntity source, final OaBrokerMainEntity target) { - if (target.getProjects().size() >= BrokerConstants.MAX_LIST_SIZE) { return new ArrayList<>(); } + if (target.getProjects().size() >= BrokerConstants.MAX_LIST_SIZE) { + return new ArrayList<>(); + } final Set existingProjects = target .getProjects() diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ClusterUtils.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ClusterUtils.java index 2d0106a7a..d8b8dd807 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ClusterUtils.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ClusterUtils.java @@ -1,6 +1,11 @@ package eu.dnetlib.dhp.broker.oa.util; +import java.util.Arrays; +import java.util.HashSet; +import java.util.Set; + +import org.apache.commons.lang3.StringUtils; import org.apache.spark.api.java.function.MapFunction; import org.apache.spark.sql.Dataset; import org.apache.spark.sql.Encoders; @@ -10,6 +15,7 @@ import org.apache.spark.util.LongAccumulator; import com.fasterxml.jackson.databind.ObjectMapper; +import eu.dnetlib.dhp.application.ArgumentApplicationParser; import eu.dnetlib.dhp.common.HdfsSupport; public class ClusterUtils { @@ -53,7 +59,9 @@ public class ClusterUtils { return o; } - public static void save(final Dataset dataset, final String path, final Class clazz, + public static void save(final Dataset dataset, + final String path, + final Class clazz, final LongAccumulator acc) { dataset .map(o -> ClusterUtils.incrementAccumulator(o, acc), Encoders.bean(clazz)) @@ -62,4 +70,20 @@ public class ClusterUtils { .json(path); } + public static Set parseParamAsList(final ArgumentApplicationParser parser, final String key) { + final String s = parser.get(key).trim(); + + final Set res = new HashSet<>(); + + if (s.length() > 1) { // A value of a single char (for example: '-') indicates an empty list + Arrays + .stream(s.split(",")) + .map(String::trim) + .filter(StringUtils::isNotBlank) + .forEach(res::add); + } + + return res; + } + } diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java index 3a2cdc5f3..d00c5b817 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/ConversionUtils.java @@ -44,7 +44,9 @@ public class ConversionUtils { private static final Logger log = LoggerFactory.getLogger(ConversionUtils.class); public static List oafInstanceToBrokerInstances(final Instance i) { - if (i == null) { return new ArrayList<>(); } + if (i == null) { + return new ArrayList<>(); + } return mappedList(i.getUrl(), url -> { final OaBrokerInstance res = new OaBrokerInstance(); @@ -65,7 +67,9 @@ public class ConversionUtils { } public static final OaBrokerRelatedDataset oafDatasetToBrokerDataset(final Dataset d) { - if (d == null) { return null; } + if (d == null) { + return null; + } final OaBrokerRelatedDataset res = new OaBrokerRelatedDataset(); res.setOpenaireId(d.getId()); @@ -78,7 +82,9 @@ public class ConversionUtils { } public static OaBrokerRelatedPublication oafPublicationToBrokerPublication(final Publication p) { - if (p == null) { return null; } + if (p == null) { + return null; + } final OaBrokerRelatedPublication res = new OaBrokerRelatedPublication(); res.setOpenaireId(p.getId()); @@ -92,7 +98,9 @@ public class ConversionUtils { } public static final OaBrokerMainEntity oafResultToBrokerResult(final Result result) { - if (result == null) { return null; } + if (result == null) { + return null; + } final OaBrokerMainEntity res = new OaBrokerMainEntity(); @@ -109,7 +117,8 @@ public class ConversionUtils { res.setEmbargoenddate(fieldValue(result.getEmbargoenddate())); res.setContributor(fieldList(result.getContributor())); res - .setJournal(result instanceof Publication ? oafJournalToBrokerJournal(((Publication) result).getJournal()) : null); + .setJournal( + result instanceof Publication ? oafJournalToBrokerJournal(((Publication) result).getJournal()) : null); res.setCollectedFromId(mappedFirst(result.getCollectedfrom(), KeyValue::getKey)); res.setCollectedFromName(mappedFirst(result.getCollectedfrom(), KeyValue::getValue)); res.setPids(mappedList(result.getPid(), ConversionUtils::oafPidToBrokerPid)); @@ -121,7 +130,9 @@ public class ConversionUtils { } private static OaBrokerAuthor oafAuthorToBrokerAuthor(final Author author) { - if (author == null) { return null; } + if (author == null) { + return null; + } final String pids = author.getPid() != null ? author .getPid() @@ -145,7 +156,9 @@ public class ConversionUtils { } private static OaBrokerJournal oafJournalToBrokerJournal(final Journal journal) { - if (journal == null) { return null; } + if (journal == null) { + return null; + } final OaBrokerJournal res = new OaBrokerJournal(); res.setName(journal.getName()); @@ -157,7 +170,9 @@ public class ConversionUtils { } private static OaBrokerExternalReference oafExtRefToBrokerExtRef(final ExternalReference ref) { - if (ref == null) { return null; } + if (ref == null) { + return null; + } final OaBrokerExternalReference res = new OaBrokerExternalReference(); res.setRefidentifier(ref.getRefidentifier()); @@ -168,7 +183,9 @@ public class ConversionUtils { } public static final OaBrokerProject oafProjectToBrokerProject(final Project p) { - if (p == null) { return null; } + if (p == null) { + return null; + } final OaBrokerProject res = new OaBrokerProject(); res.setOpenaireId(p.getId()); @@ -192,7 +209,9 @@ public class ConversionUtils { } public static final OaBrokerRelatedSoftware oafSoftwareToBrokerSoftware(final Software sw) { - if (sw == null) { return null; } + if (sw == null) { + return null; + } final OaBrokerRelatedSoftware res = new OaBrokerRelatedSoftware(); res.setOpenaireId(sw.getId()); @@ -255,7 +274,9 @@ public class ConversionUtils { } private static List structPropTypedList(final List list) { - if (list == null) { return new ArrayList<>(); } + if (list == null) { + return new ArrayList<>(); + } return list .stream() @@ -265,7 +286,9 @@ public class ConversionUtils { } private static List mappedList(final List list, final Function func) { - if (list == null) { return new ArrayList<>(); } + if (list == null) { + return new ArrayList<>(); + } return list .stream() @@ -276,7 +299,9 @@ public class ConversionUtils { } private static List flatMappedList(final List list, final Function> func) { - if (list == null) { return new ArrayList<>(); } + if (list == null) { + return new ArrayList<>(); + } return list .stream() @@ -288,7 +313,9 @@ public class ConversionUtils { } private static T mappedFirst(final List list, final Function func) { - if (list == null) { return null; } + if (list == null) { + return null; + } return list .stream() diff --git a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/EventFinder.java b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/EventFinder.java index e7abae68b..593e66d43 100644 --- a/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/EventFinder.java +++ b/dhp-workflows/dhp-broker-events/src/main/java/eu/dnetlib/dhp/broker/oa/util/EventFinder.java @@ -70,7 +70,6 @@ public class EventFinder { matchers.add(new EnrichMissingDatasetReferences()); matchers.add(new EnrichMissingDatasetIsSupplementedTo()); matchers.add(new EnrichMissingDatasetIsSupplementedBy()); - matchers.add(new EnrichMissingAbstract()); } public static EventGroup generateEvents(final ResultGroup results, diff --git a/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/generate_all/oozie_app/workflow.xml b/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/generate_all/oozie_app/workflow.xml index a0c7b00db..b85c60fdf 100644 --- a/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/generate_all/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/generate_all/oozie_app/workflow.xml @@ -447,7 +447,7 @@ --executor-memory=${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} - --conf spark.dynamicAllocation.maxExecutors="2" + --conf spark.dynamicAllocation.maxExecutors="8" --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} diff --git a/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/partial/oozie_app/workflow.xml b/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/partial/oozie_app/workflow.xml index b38290448..392271260 100644 --- a/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/partial/oozie_app/workflow.xml +++ b/dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/partial/oozie_app/workflow.xml @@ -73,33 +73,32 @@ - + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - + yarn cluster - IndexOnESJob - eu.dnetlib.dhp.broker.oa.IndexOnESJob + Count + eu.dnetlib.dhp.broker.oa.CheckDuplictedIdsJob dhp-broker-events-${projectVersion}.jar + --executor-cores=${sparkExecutorCores} --executor-memory=${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} - --conf spark.dynamicAllocation.maxExecutors="2" --conf spark.extraListeners=${spark2ExtraListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.sql.shuffle.partitions=3840 + --graphPath${graphInputPath} --workingPath${workingPath} - --index${esIndexName} - --esHost${esIndexHost}