Fixed problem on duplicated identifier

This commit is contained in:
Sandro La Bruzzo 2020-11-25 17:15:54 +01:00
parent 264723ffd8
commit ec3e238de6
3 changed files with 116 additions and 95 deletions

View File

@ -14,7 +14,7 @@ import scala.collection.JavaConverters._
import scala.collection.mutable
import scala.util.matching.Regex
case class CrossrefDT(doi: String, json:String) {}
case class CrossrefDT(doi: String, json:String, timestamp: Long) {}
case class mappingAffiliation(name: String) {}

View File

@ -29,69 +29,90 @@ object SparkMapDumpIntoOAF {
.appName(SparkMapDumpIntoOAF.getClass.getSimpleName)
.master(parser.get("master")).getOrCreate()
implicit val oafEncoder: Encoder[Oaf] = Encoders.kryo[Oaf]
implicit val mapEncoderPubs: Encoder[Publication] = Encoders.kryo[Publication]
implicit val mapEncoderRelatons: Encoder[Relation] = Encoders.kryo[Relation]
implicit val mapEncoderDatasets: Encoder[oaf.Dataset] = Encoders.kryo[OafDataset]
val sc = spark.sparkContext
val targetPath = parser.get("targetPath")
import spark.implicits._
sc.sequenceFile(parser.get("sourcePath"), classOf[IntWritable], classOf[Text])
.map(k => k._2.toString).map(CrossrefImporter.decompressBlob)
.flatMap(k => Crossref2Oaf.convert(k)).saveAsObjectFile(s"${targetPath}/mixObject")
val inputRDD = sc.objectFile[Oaf](s"${targetPath}/mixObject").filter(p=> p!= null)
val distinctPubs:RDD[Publication] = inputRDD.filter(k => k != null && k.isInstanceOf[Publication])
.map(k => k.asInstanceOf[Publication]).map { p: Publication => Tuple2(p.getId, p) }.reduceByKey { case (p1: Publication, p2: Publication) =>
var r = if (p1 == null) p2 else p1
if (p1 != null && p2 != null) {
if (p1.getLastupdatetimestamp != null && p2.getLastupdatetimestamp != null) {
if (p1.getLastupdatetimestamp < p2.getLastupdatetimestamp)
r = p2
else
r = p1
} else {
r = if (p1.getLastupdatetimestamp == null) p2 else p1
}
}
r
}.map(_._2)
val pubs:Dataset[Publication] = spark.createDataset(distinctPubs)
pubs.write.mode(SaveMode.Overwrite).save(s"${targetPath}/publication")
spark.read.load(parser.get("sourcePath")).as[CrossrefDT]
.flatMap(k => Crossref2Oaf.convert(k.json))
.filter(o => o != null)
.write.mode(SaveMode.Overwrite).save(s"$targetPath/mixObject")
val distincDatasets:RDD[OafDataset] = inputRDD.filter(k => k != null && k.isInstanceOf[OafDataset])
.map(k => k.asInstanceOf[OafDataset]).map(p => Tuple2(p.getId, p)).reduceByKey { case (p1: OafDataset, p2: OafDataset) =>
var r = if (p1 == null) p2 else p1
if (p1 != null && p2 != null) {
if (p1.getLastupdatetimestamp != null && p2.getLastupdatetimestamp != null) {
if (p1.getLastupdatetimestamp < p2.getLastupdatetimestamp)
r = p2
else
r = p1
} else {
r = if (p1.getLastupdatetimestamp == null) p2 else p1
}
}
r
}.map(_._2)
val ds:Dataset[Oaf] = spark.read.load(s"$targetPath/mixObject").as[Oaf]
spark.createDataset(distincDatasets).write.mode(SaveMode.Overwrite).save(s"${targetPath}/dataset")
ds.filter(o => o.isInstanceOf[Publication]).map(o => o.asInstanceOf[Publication]).write.save(s"$targetPath/publication")
ds.filter(o => o.isInstanceOf[Relation]).map(o => o.asInstanceOf[Relation]).write.save(s"$targetPath/relation")
ds.filter(o => o.isInstanceOf[OafDataset]).map(o => o.asInstanceOf[OafDataset]).write.save(s"$targetPath/dataset")
val distinctRels =inputRDD.filter(k => k != null && k.isInstanceOf[Relation])
.map(k => k.asInstanceOf[Relation]).map(r=> (s"${r.getSource}::${r.getTarget}",r))
.reduceByKey { case (p1: Relation, p2: Relation) =>
if (p1 == null) p2 else p1
}.map(_._2)
val rels: Dataset[Relation] = spark.createDataset(distinctRels)
rels.write.mode(SaveMode.Overwrite).save(s"${targetPath}/relations")
//
//
//
// sc.sequenceFile(parser.get("sourcePath"), classOf[IntWritable], classOf[Text])
// .map(k => k._2.toString).map(CrossrefImporter.decompressBlob)
// .flatMap(k => Crossref2Oaf.convert(k)).saveAsObjectFile(s"${targetPath}/mixObject")
//
// val inputRDD = sc.objectFile[Oaf](s"${targetPath}/mixObject").filter(p=> p!= null)
//
// val distinctPubs:RDD[Publication] = inputRDD.filter(k => k != null && k.isInstanceOf[Publication])
// .map(k => k.asInstanceOf[Publication]).map { p: Publication => Tuple2(p.getId, p) }.reduceByKey { case (p1: Publication, p2: Publication) =>
// var r = if (p1 == null) p2 else p1
// if (p1 != null && p2 != null) {
// if (p1.getLastupdatetimestamp != null && p2.getLastupdatetimestamp != null) {
// if (p1.getLastupdatetimestamp < p2.getLastupdatetimestamp)
// r = p2
// else
// r = p1
// } else {
// r = if (p1.getLastupdatetimestamp == null) p2 else p1
// }
// }
// r
// }.map(_._2)
//
// val pubs:Dataset[Publication] = spark.createDataset(distinctPubs)
// pubs.write.mode(SaveMode.Overwrite).save(s"${targetPath}/publication")
//
//
// val distincDatasets:RDD[OafDataset] = inputRDD.filter(k => k != null && k.isInstanceOf[OafDataset])
// .map(k => k.asInstanceOf[OafDataset]).map(p => Tuple2(p.getId, p)).reduceByKey { case (p1: OafDataset, p2: OafDataset) =>
// var r = if (p1 == null) p2 else p1
// if (p1 != null && p2 != null) {
// if (p1.getLastupdatetimestamp != null && p2.getLastupdatetimestamp != null) {
// if (p1.getLastupdatetimestamp < p2.getLastupdatetimestamp)
// r = p2
// else
// r = p1
// } else {
// r = if (p1.getLastupdatetimestamp == null) p2 else p1
// }
// }
// r
// }.map(_._2)
//
// spark.createDataset(distincDatasets).write.mode(SaveMode.Overwrite).save(s"${targetPath}/dataset")
//
//
//
// val distinctRels =inputRDD.filter(k => k != null && k.isInstanceOf[Relation])
// .map(k => k.asInstanceOf[Relation]).map(r=> (s"${r.getSource}::${r.getTarget}",r))
// .reduceByKey { case (p1: Relation, p2: Relation) =>
// if (p1 == null) p2 else p1
// }.map(_._2)
//
// val rels: Dataset[Relation] = spark.createDataset(distinctRels)
//
// rels.write.mode(SaveMode.Overwrite).save(s"${targetPath}/relations")
}

View File

@ -16,10 +16,10 @@
<name>sparkExecutorCores</name>
<description>number of cores used by single executor</description>
</property>
<property>
<name>timestamp</name>
<description>Timestamp for incremental Harvesting</description>
</property>
<!-- <property>-->
<!-- <name>timestamp</name>-->
<!-- <description>Timestamp for incremental Harvesting</description>-->
<!-- </property>-->
</parameters>
@ -30,29 +30,29 @@
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="ResetWorkingPath">
<fs>
<delete path='${workingPath}/input/crossref/index_dump'/>
<!-- <mkdir path='${workingPath}/input/crossref'/>-->
</fs>
<ok to="ImportCrossRef"/>
<error to="Kill"/>
</action>
<!-- <action name="ResetWorkingPath">-->
<!-- <fs>-->
<!-- <delete path='${workingPath}/input/crossref/index_dump'/>-->
<!--&lt;!&ndash; <mkdir path='${workingPath}/input/crossref'/>&ndash;&gt;-->
<!-- </fs>-->
<!-- <ok to="ImportCrossRef"/>-->
<!-- <error to="Kill"/>-->
<!-- </action>-->
<action name="ImportCrossRef">
<java>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<main-class>eu.dnetlib.doiboost.crossref.CrossrefImporter</main-class>
<arg>-t</arg><arg>${workingPath}/input/crossref/index_dump_1</arg>
<arg>-n</arg><arg>${nameNode}</arg>
<arg>-ts</arg><arg>${timestamp}</arg>
</java>
<ok to="End"/>
<error to="Kill"/>
</action>
<!-- <action name="ImportCrossRef">-->
<!-- <java>-->
<!-- <job-tracker>${jobTracker}</job-tracker>-->
<!-- <name-node>${nameNode}</name-node>-->
<!-- <main-class>eu.dnetlib.doiboost.crossref.CrossrefImporter</main-class>-->
<!-- <arg>-t</arg><arg>${workingPath}/input/crossref/index_dump_1</arg>-->
<!-- <arg>-n</arg><arg>${nameNode}</arg>-->
<!-- <arg>-ts</arg><arg>${timestamp}</arg>-->
<!-- </java>-->
<!-- <ok to="End"/>-->
<!-- <error to="Kill"/>-->
<!-- </action>-->
<action name="ExtractCrossrefToOAF">
@ -68,7 +68,7 @@
--driver-memory=${sparkDriverMemory}
${sparkExtraOPT}
</spark-opts>
<arg>--sourcePath</arg><arg>${workingPath}/input/crossref/index_dump,${workingPath}/input/crossref/index_dump_1,${workingPath}/crossref/index_dump</arg>
<arg>--sourcePath</arg><arg>${workingPath}/input/crossref/crossref_ds</arg>
<arg>--targetPath</arg><arg>${workingPath}/input/crossref</arg>
<arg>--master</arg><arg>yarn-cluster</arg>
</spark>
@ -78,26 +78,26 @@
<action name="GenerateDataset">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>ExtractCrossrefToOAF</name>
<class>eu.dnetlib.doiboost.crossref.CrossrefDataset</class>
<jar>dhp-doiboost-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
${sparkExtraOPT}
</spark-opts>
<arg>--sourcePath</arg><arg>/data/doiboost/crossref/cr_dataset</arg>
<arg>--targetPath</arg><arg>/data/doiboost/crossref/crossrefDataset</arg>
<arg>--master</arg><arg>yarn-cluster</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<!-- <action name="GenerateDataset">-->
<!-- <spark xmlns="uri:oozie:spark-action:0.2">-->
<!-- <master>yarn-cluster</master>-->
<!-- <mode>cluster</mode>-->
<!-- <name>ExtractCrossrefToOAF</name>-->
<!-- <class>eu.dnetlib.doiboost.crossref.CrossrefDataset</class>-->
<!-- <jar>dhp-doiboost-${projectVersion}.jar</jar>-->
<!-- <spark-opts>-->
<!-- &#45;&#45;executor-memory=${sparkExecutorMemory}-->
<!-- &#45;&#45;executor-cores=${sparkExecutorCores}-->
<!-- &#45;&#45;driver-memory=${sparkDriverMemory}-->
<!-- ${sparkExtraOPT}-->
<!-- </spark-opts>-->
<!-- <arg>&#45;&#45;sourcePath</arg><arg>/data/doiboost/crossref/cr_dataset</arg>-->
<!-- <arg>&#45;&#45;targetPath</arg><arg>/data/doiboost/crossref/crossrefDataset</arg>-->
<!-- <arg>&#45;&#45;master</arg><arg>yarn-cluster</arg>-->
<!-- </spark>-->
<!-- <ok to="End"/>-->
<!-- <error to="Kill"/>-->
<!-- </action>-->
<end name="End"/>
</workflow-app>