mergin with branch beta

This commit is contained in:
Miriam Baglioni 2021-07-28 11:05:16 +02:00
commit 73ed7374a9
8 changed files with 295 additions and 6 deletions

View File

@ -1,9 +1,10 @@
package eu.dnetlib.dhp.actionmanager.datacite package eu.dnetlib.dhp.actionmanager.datacite
import org.apache.commons.io.IOUtils import org.apache.commons.io.IOUtils
import org.apache.http.client.config.RequestConfig
import org.apache.http.client.methods.{HttpGet, HttpPost, HttpRequestBase, HttpUriRequest} import org.apache.http.client.methods.{HttpGet, HttpPost, HttpRequestBase, HttpUriRequest}
import org.apache.http.entity.StringEntity import org.apache.http.entity.StringEntity
import org.apache.http.impl.client.HttpClients import org.apache.http.impl.client.{HttpClientBuilder, HttpClients}
import java.io.IOException import java.io.IOException
@ -56,11 +57,15 @@ abstract class AbstractRestClient extends Iterator[String]{
private def doHTTPRequest[A <: HttpUriRequest](r: A) :String ={ private def doHTTPRequest[A <: HttpUriRequest](r: A) :String ={
val client = HttpClients.createDefault val timeout = 60; // seconds
val config = RequestConfig.custom()
.setConnectTimeout(timeout * 1000)
.setConnectionRequestTimeout(timeout * 1000)
.setSocketTimeout(timeout * 1000).build()
val client =HttpClientBuilder.create().setDefaultRequestConfig(config).build()
var tries = 4 var tries = 4
try { try {
while (tries > 0) { while (tries > 0) {
println(s"requesting ${r.getURI}") println(s"requesting ${r.getURI}")
val response = client.execute(r) val response = client.execute(r)
println(s"get response with status${response.getStatusLine.getStatusCode}") println(s"get response with status${response.getStatusLine.getStatusCode}")
@ -80,7 +85,5 @@ abstract class AbstractRestClient extends Iterator[String]{
throw new RuntimeException("Unable to close client ", e) throw new RuntimeException("Unable to close client ", e)
} }
} }
getBufferData() getBufferData()
} }

View File

@ -140,7 +140,7 @@ object ImportDatacite {
private def writeSequenceFile(hdfsTargetPath: Path, timestamp: Long, conf: Configuration, bs:Int): Long = { private def writeSequenceFile(hdfsTargetPath: Path, timestamp: Long, conf: Configuration, bs:Int): Long = {
var from:Long = timestamp * 1000 var from:Long = timestamp * 1000
val delta:Long = 50000000L val delta:Long = 100000000L
var client: DataciteAPIImporter = null var client: DataciteAPIImporter = null
val now :Long =System.currentTimeMillis() val now :Long =System.currentTimeMillis()
var i = 0 var i = 0

View File

@ -0,0 +1,90 @@
package eu.dnetlib.dhp.sx.provision
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.oaf.{Oaf, Relation, Result}
import org.apache.spark.{SparkConf, sql}
import org.apache.spark.sql.{Dataset, Encoder, Encoders, SaveMode, SparkSession}
import org.slf4j.{Logger, LoggerFactory}
import scala.io.Source
object SparkCreateActionset {
def main(args: Array[String]): Unit = {
val log: Logger = LoggerFactory.getLogger(getClass)
val conf: SparkConf = new SparkConf()
val parser = new ArgumentApplicationParser(Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/actionset/generate_actionset.json")).mkString)
parser.parseArgument(args)
val spark: SparkSession =
SparkSession
.builder()
.config(conf)
.appName(getClass.getSimpleName)
.master(parser.get("master")).getOrCreate()
val sourcePath = parser.get("sourcePath")
log.info(s"sourcePath -> $sourcePath")
val targetPath = parser.get("targetPath")
log.info(s"targetPath -> $targetPath")
val workingDirFolder = parser.get("workingDirFolder")
log.info(s"workingDirFolder -> $workingDirFolder")
implicit val oafEncoders:Encoder[Oaf] = Encoders.kryo[Oaf]
implicit val resultEncoders:Encoder[Result] = Encoders.kryo[Result]
implicit val relationEncoders:Encoder[Relation] = Encoders.kryo[Relation]
import spark.implicits._
val relation = spark.read.load(s"$sourcePath/relation").as[Relation]
relation.filter(r => (r.getDataInfo== null || r.getDataInfo.getDeletedbyinference == false) && !r.getRelClass.toLowerCase.contains("merge"))
.flatMap(r => List(r.getSource,r.getTarget)).distinct().write.save(s"$workingDirFolder/id_relation")
val idRelation = spark.read.load(s"$workingDirFolder/id_relation").as[String]
log.info("extract source and target Identifier involved in relations")
log.info("save relation filtered")
relation.filter(r => (r.getDataInfo== null || r.getDataInfo.getDeletedbyinference == false) && !r.getRelClass.toLowerCase.contains("merge"))
.write.mode(SaveMode.Overwrite).save(s"$workingDirFolder/actionSetOaf")
log.info("saving publication")
val publication:Dataset[(String, Result)] = spark.read.load(s"$sourcePath/publication").as[Result].map(p => (p.getId, p))
publication
.joinWith(idRelation, publication("_1").equalTo(idRelation("value")))
.map(p => p._1._2)
.write.mode(SaveMode.Append).save(s"$workingDirFolder/actionSetOaf")
log.info("saving dataset")
val dataset:Dataset[(String, Result)] = spark.read.load(s"$sourcePath/dataset").as[Result].map(p => (p.getId, p))
dataset
.joinWith(idRelation, publication("_1").equalTo(idRelation("value")))
.map(p => p._1._2)
.write.mode(SaveMode.Append).save(s"$workingDirFolder/actionSetOaf")
log.info("saving software")
val software:Dataset[(String, Result)] = spark.read.load(s"$sourcePath/software").as[Result].map(p => (p.getId, p))
software
.joinWith(idRelation, publication("_1").equalTo(idRelation("value")))
.map(p => p._1._2)
.write.mode(SaveMode.Append).save(s"$workingDirFolder/actionSetOaf")
log.info("saving Other Research product")
val orp:Dataset[(String, Result)] = spark.read.load(s"$sourcePath/otherresearchproduct").as[Result].map(p => (p.getId, p))
orp
.joinWith(idRelation, publication("_1").equalTo(idRelation("value")))
.map(p => p._1._2)
.write.mode(SaveMode.Append).save(s"$workingDirFolder/actionSetOaf")
}
}

View File

@ -0,0 +1,86 @@
package eu.dnetlib.dhp.sx.provision
import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.action.AtomicAction
import eu.dnetlib.dhp.schema.oaf.{Oaf, OtherResearchProduct, Publication, Relation, Software, Dataset => OafDataset}
import org.apache.hadoop.io.Text
import org.apache.hadoop.io.compress.GzipCodec
import org.apache.hadoop.mapred.SequenceFileOutputFormat
import org.apache.spark.SparkConf
import org.apache.spark.sql.{Encoder, Encoders, SparkSession}
import org.slf4j.{Logger, LoggerFactory}
import scala.io.Source
object SparkSaveActionSet {
def toActionSet(item: Oaf): (String, String) = {
val mapper = new ObjectMapper()
item match {
case dataset: OafDataset =>
val a: AtomicAction[OafDataset] = new AtomicAction[OafDataset]
a.setClazz(classOf[OafDataset])
a.setPayload(dataset)
(dataset.getClass.getCanonicalName, mapper.writeValueAsString(a))
case publication: Publication =>
val a: AtomicAction[Publication] = new AtomicAction[Publication]
a.setClazz(classOf[Publication])
a.setPayload(publication)
(publication.getClass.getCanonicalName, mapper.writeValueAsString(a))
case software: Software =>
val a: AtomicAction[Software] = new AtomicAction[Software]
a.setClazz(classOf[Software])
a.setPayload(software)
(software.getClass.getCanonicalName, mapper.writeValueAsString(a))
case orp: OtherResearchProduct =>
val a: AtomicAction[OtherResearchProduct] = new AtomicAction[OtherResearchProduct]
a.setClazz(classOf[OtherResearchProduct])
a.setPayload(orp)
(orp.getClass.getCanonicalName, mapper.writeValueAsString(a))
case relation: Relation =>
val a: AtomicAction[Relation] = new AtomicAction[Relation]
a.setClazz(classOf[Relation])
a.setPayload(relation)
(relation.getClass.getCanonicalName, mapper.writeValueAsString(a))
case _ =>
null
}
}
def main(args: Array[String]): Unit = {
val log: Logger = LoggerFactory.getLogger(getClass)
val conf: SparkConf = new SparkConf()
val parser = new ArgumentApplicationParser(Source.fromInputStream(getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/actionset/save_actionset.json")).mkString)
parser.parseArgument(args)
val spark: SparkSession =
SparkSession
.builder()
.config(conf)
.appName(getClass.getSimpleName)
.master(parser.get("master")).getOrCreate()
val sourcePath = parser.get("sourcePath")
log.info(s"sourcePath -> $sourcePath")
val targetPath = parser.get("targetPath")
log.info(s"targetPath -> $targetPath")
implicit val oafEncoders:Encoder[Oaf] = Encoders.kryo[Oaf]
implicit val tEncoder:Encoder[(String,String)] = Encoders.tuple(Encoders.STRING,Encoders.STRING)
spark.read.load(sourcePath).as[Oaf]
.map(o =>toActionSet(o))
.filter(o => o!= null)
.rdd.map(s => (new Text(s._1), new Text(s._2))).saveAsHadoopFile(s"$targetPath", classOf[Text], classOf[Text], classOf[SequenceFileOutputFormat[Text,Text]], classOf[GzipCodec])
}
}

View File

@ -0,0 +1,6 @@
[
{"paramName":"mt", "paramLongName":"master", "paramDescription": "should be local or yarn", "paramRequired": true},
{"paramName":"s", "paramLongName":"sourcePath","paramDescription": "source path", "paramRequired": true},
{"paramName":"w", "paramLongName":"workingDirFolder","paramDescription": "the working Dir Folder", "paramRequired": true},
{"paramName":"t", "paramLongName":"targetPath","paramDescription": "the target path ", "paramRequired": true}
]

View File

@ -0,0 +1,23 @@
<configuration>
<property>
<name>jobTracker</name>
<value>yarnRM</value>
</property>
<property>
<name>nameNode</name>
<value>hdfs://nameservice1</value>
</property>
<property>
<name>oozie.use.system.libpath</name>
<value>true</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>spark2</value>
</property>
<property>
<name>oozie.launcher.mapreduce.user.classpath.first</name>
<value>true</value>
</property>
</configuration>

View File

@ -0,0 +1,76 @@
<workflow-app name="Scholexplorer_to_ActionSet_Workflow" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>sourcePath</name>
<description>the path of the consistent graph</description>
</property>
<property>
<name>workingDirFolder</name>
<description>the path of working dir ActionSet</description>
</property>
<property>
<name>outputPath</name>
<description>the path of Scholexplorer ActionSet</description>
</property>
</parameters>
<start to="ExportDataset"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="createActionSet">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>Create Action Set</name>
<class>eu.dnetlib.dhp.sx.provision.SparkCreateActionset</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.sql.shuffle.partitions=3840
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
<arg>--targetPath</arg><arg>${outputPath}</arg>
<arg>--workingDirFolder</arg><arg>${workingDirFolder}</arg>
<arg>--master</arg><arg>yarn-cluster</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SaveActionSet">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>Save Action Set</name>
<class>eu.dnetlib.dhp.sx.provision.SparkSaveActionSet</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.sql.shuffle.partitions=3840
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--sourcePath</arg><arg>${workingDirFolder}/actionSetOaf</arg>
<arg>--targetPath</arg><arg>${outputPath}</arg>
<arg>--master</arg><arg>yarn-cluster</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

View File

@ -0,0 +1,5 @@
[
{"paramName":"mt", "paramLongName":"master", "paramDescription": "should be local or yarn", "paramRequired": true},
{"paramName":"s", "paramLongName":"sourcePath","paramDescription": "source path", "paramRequired": true},
{"paramName":"t", "paramLongName":"targetPath","paramDescription": "the target path ", "paramRequired": true}
]