Merge pull request 'implemented oozie workflow to generate scholix dump filtering relclass semantic' (#229) from opencitation_enrichments into beta

Reviewed-on: D-Net/dnet-hadoop#229
This commit is contained in:
Claudio Atzori 2022-07-21 10:12:17 +02:00
commit d900a02b74
6 changed files with 226 additions and 26 deletions

View File

@ -2,4 +2,5 @@
{"paramName":"mt", "paramLongName":"master", "paramDescription": "should be local or yarn", "paramRequired": true},
{"paramName":"s", "paramLongName":"sourcePath", "paramDescription": "the source Path", "paramRequired": true},
{"paramName":"t", "paramLongName":"targetPath", "paramDescription": "the path of the raw graph", "paramRequired": true}
{"paramName":"r", "paramLongName":"filterRelation", "paramDescription": "the relation to filter", "paramRequired": false}
]

View File

@ -3,5 +3,7 @@
{"paramName":"s", "paramLongName":"sourcePath", "paramDescription": "the source Path", "paramRequired": true},
{"paramName":"su", "paramLongName":"scholixUpdatePath", "paramDescription": "the scholix updated Path", "paramRequired": false},
{"paramName":"t", "paramLongName":"targetPath", "paramDescription": "the path of the raw graph", "paramRequired": true},
{"paramName":"o", "paramLongName":"objectType", "paramDescription": "should be scholix or Summary", "paramRequired": true}
{"paramName":"o", "paramLongName":"objectType", "paramDescription": "should be scholix or Summary", "paramRequired": true},
{"paramName":"mp", "paramLongName":"maxPidNumberFilter", "paramDescription": "filter max number of pids in source/target", "paramRequired": false}
]

View File

@ -0,0 +1,10 @@
<configuration>
<property>
<name>oozie.use.system.libpath</name>
<value>true</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>spark2</value>
</property>
</configuration>

View File

@ -0,0 +1,145 @@
<workflow-app name="Create Scholix Dump" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>sourcePath</name>
<description>the working dir base path</description>
</property>
<property>
<name>targetPath</name>
<description>the final graph path</description>
</property>
<property>
<name>relationFilter</name>
<description>Filter relation semantic</description>
</property>
<property>
<name>maxNumberOfPid</name>
<description>filter relation with at least #maxNumberOfPid</description>
</property>
</parameters>
<start to="ImportDatasetEntities"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="ImportDatasetEntities">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Import JSONRDD to Dataset kryo</name>
<class>eu.dnetlib.dhp.sx.graph.SparkConvertRDDtoDataset</class>
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.shuffle.partitions=3000
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn</arg>
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
<arg>--targetPath</arg><arg>${targetPath}</arg>
<arg>--filterRelation</arg><arg>${relationFilter}</arg>
</spark>
<ok to="CreateSummaries"/>
<error to="Kill"/>
</action>
<action name="CreateSummaries">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Convert Entities to summaries</name>
<class>eu.dnetlib.dhp.sx.graph.SparkCreateSummaryObject</class>
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.shuffle.partitions=20000
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn</arg>
<arg>--sourcePath</arg><arg>${targetPath}/entities</arg>
<arg>--targetPath</arg><arg>${targetPath}/provision/summaries</arg>
</spark>
<ok to="CreateScholix"/>
<error to="Kill"/>
</action>
<action name="CreateScholix">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Generate Scholix Dataset</name>
<class>eu.dnetlib.dhp.sx.graph.SparkCreateScholix</class>
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.shuffle.partitions=30000
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn</arg>
<arg>--summaryPath</arg><arg>${targetPath}/provision/summaries</arg>
<arg>--targetPath</arg><arg>${targetPath}/provision/scholix</arg>
<arg>--relationPath</arg><arg>${targetPath}/relation</arg>
</spark>
<ok to="DropJSONPath"/>
<error to="Kill"/>
</action>
<action name="DropJSONPath">
<fs>
<delete path='${targetPath}/json'/>
<mkdir path='${targetPath}/json/'/>
</fs>
<ok to="SerializeScholix"/>
<error to="Kill"/>
</action>
<action name="SerializeScholix">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Serialize scholix to JSON</name>
<class>eu.dnetlib.dhp.sx.graph.SparkConvertObjectToJson</class>
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.shuffle.partitions=6000
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn</arg>
<arg>--sourcePath</arg><arg>${targetPath}/provision/scholix/scholix</arg>
<arg>--targetPath</arg><arg>${targetPath}/json/scholix_json</arg>
<arg>--objectType</arg><arg>scholix</arg>
<arg>--maxPidNumberFilter</arg><arg>maxNumberOfPid</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

View File

@ -4,6 +4,7 @@ import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.sx.scholix.Scholix
import eu.dnetlib.dhp.schema.sx.summary.ScholixSummary
import eu.dnetlib.dhp.sx.graph.SparkConvertObjectToJson.toInt
import org.apache.commons.io.IOUtils
import org.apache.hadoop.io.compress.GzipCodec
import org.apache.spark.SparkConf
@ -12,6 +13,14 @@ import org.slf4j.{Logger, LoggerFactory}
object SparkConvertObjectToJson {
def toInt(s: String): Option[Int] = {
try {
Some(s.toInt)
} catch {
case e: Exception => None
}
}
def main(args: Array[String]): Unit = {
val log: Logger = LoggerFactory.getLogger(getClass)
val conf: SparkConf = new SparkConf()
@ -37,6 +46,8 @@ object SparkConvertObjectToJson {
log.info(s"objectType -> $objectType")
val scholixUpdatePath = parser.get("scholixUpdatePath")
log.info(s"scholixUpdatePath -> $scholixUpdatePath")
val maxPidNumberFilter = parser.get("maxPidNumberFilter")
log.info(s"maxPidNumberFilter -> $maxPidNumberFilter")
implicit val scholixEncoder: Encoder[Scholix] = Encoders.kryo[Scholix]
implicit val summaryEncoder: Encoder[ScholixSummary] = Encoders.kryo[ScholixSummary]
@ -47,12 +58,22 @@ object SparkConvertObjectToJson {
case "scholix" =>
log.info("Serialize Scholix")
val d: Dataset[Scholix] = spark.read.load(sourcePath).as[Scholix]
val u: Dataset[Scholix] = spark.read.load(s"$scholixUpdatePath/scholix").as[Scholix]
d.union(u)
.repartition(8000)
.map(s => mapper.writeValueAsString(s))(Encoders.STRING)
.rdd
.saveAsTextFile(targetPath, classOf[GzipCodec])
// val u: Dataset[Scholix] = spark.read.load(s"$scholixUpdatePath/scholix").as[Scholix]
if (maxPidNumberFilter != null && toInt(maxPidNumberFilter).isDefined) {
val mp = toInt(maxPidNumberFilter).get
d
.filter(s => (s.getSource.getIdentifier.size() <= mp) && (s.getTarget.getIdentifier.size() <= mp))
.map(s => mapper.writeValueAsString(s))(Encoders.STRING)
.rdd
.saveAsTextFile(targetPath, classOf[GzipCodec])
} else {
d
.repartition(8000)
.map(s => mapper.writeValueAsString(s))(Encoders.STRING)
.rdd
.saveAsTextFile(targetPath, classOf[GzipCodec])
}
case "summary" =>
log.info("Serialize Summary")
val d: Dataset[ScholixSummary] = spark.read.load(sourcePath).as[ScholixSummary]

View File

@ -4,9 +4,11 @@ import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.oaf.{OtherResearchProduct, Publication, Relation, Result, Software, Dataset => OafDataset}
import org.apache.commons.io.IOUtils
import org.apache.commons.lang3.StringUtils
import org.apache.spark.SparkConf
import org.apache.spark.sql.{Encoder, Encoders, SaveMode, SparkSession}
import org.slf4j.{Logger, LoggerFactory}
import scala.collection.JavaConverters._
object SparkConvertRDDtoDataset {
@ -34,6 +36,9 @@ object SparkConvertRDDtoDataset {
val t = parser.get("targetPath")
log.info(s"targetPath -> $t")
val filterRelation = parser.get("filterRelation")
log.info(s"filterRelation -> $filterRelation")
val entityPath = s"$t/entities"
val relPath = s"$t/relation"
val mapper = new ObjectMapper()
@ -94,28 +99,44 @@ object SparkConvertRDDtoDataset {
log.info("Converting Relation")
val relationSemanticFilter = List(
// "cites",
// "iscitedby",
"merges",
"ismergedin",
"HasAmongTopNSimilarDocuments",
"IsAmongTopNSimilarDocuments"
)
if (filterRelation != null && StringUtils.isNoneBlank(filterRelation)) {
val rddRelation = spark.sparkContext
.textFile(s"$sourcePath/relation")
.map(s => mapper.readValue(s, classOf[Relation]))
.filter(r => r.getDataInfo != null && r.getDataInfo.getDeletedbyinference == false)
.filter(r => r.getSource.startsWith("50") && r.getTarget.startsWith("50"))
//filter OpenCitations relations
.filter(r =>
r.getCollectedfrom != null && r.getCollectedfrom.size() > 0 && !r.getCollectedfrom.asScala.exists(k =>
"opencitations".equalsIgnoreCase(k.getValue)
val rddRelation = spark.sparkContext
.textFile(s"$sourcePath/relation")
.map(s => mapper.readValue(s, classOf[Relation]))
.filter(r => r.getDataInfo != null && r.getDataInfo.getDeletedbyinference == false)
.filter(r => r.getSource.startsWith("50") && r.getTarget.startsWith("50"))
//filter OpenCitations relations
.filter(r =>
r.getCollectedfrom != null && r.getCollectedfrom.size() > 0 && !r.getCollectedfrom.asScala.exists(k =>
"opencitations".equalsIgnoreCase(k.getValue)
)
)
.filter(r => r.getSubRelType != null && r.getSubRelType.equalsIgnoreCase(filterRelation))
spark.createDataset(rddRelation).as[Relation].write.mode(SaveMode.Overwrite).save(s"$relPath")
} else {
val relationSemanticFilter = List(
"merges",
"ismergedin",
"HasAmongTopNSimilarDocuments",
"IsAmongTopNSimilarDocuments"
)
.filter(r => !relationSemanticFilter.exists(k => k.equalsIgnoreCase(r.getRelClass)))
spark.createDataset(rddRelation).as[Relation].write.mode(SaveMode.Overwrite).save(s"$relPath")
val rddRelation = spark.sparkContext
.textFile(s"$sourcePath/relation")
.map(s => mapper.readValue(s, classOf[Relation]))
.filter(r => r.getDataInfo != null && r.getDataInfo.getDeletedbyinference == false)
.filter(r => r.getSource.startsWith("50") && r.getTarget.startsWith("50"))
//filter OpenCitations relations
.filter(r =>
r.getCollectedfrom != null && r.getCollectedfrom.size() > 0 && !r.getCollectedfrom.asScala.exists(k =>
"opencitations".equalsIgnoreCase(k.getValue)
)
)
.filter(r => !relationSemanticFilter.exists(k => k.equalsIgnoreCase(r.getRelClass)))
spark.createDataset(rddRelation).as[Relation].write.mode(SaveMode.Overwrite).save(s"$relPath")
}
}
}