2021-08-13 16:39:02 +02:00
|
|
|
<workflow-app name="Generate DOIBoost ActionSet - PROCESS" xmlns="uri:oozie:workflow:0.5">
|
2021-01-04 17:01:35 +01:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>sparkDriverMemory</name>
|
|
|
|
<description>memory for driver process</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>sparkExecutorMemory</name>
|
|
|
|
<description>memory for individual executor</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>sparkExecutorIntersectionMemory</name>
|
|
|
|
<description>memory for individual executor</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>sparkExecutorCores</name>
|
|
|
|
<description>number of cores used by single executor</description>
|
|
|
|
</property>
|
|
|
|
<!-- Itersection Parameters -->
|
|
|
|
<property>
|
|
|
|
<name>workingPath</name>
|
|
|
|
<description>the working Path</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>hostedByMapPath</name>
|
|
|
|
<description>the hostedByMap Path</description>
|
|
|
|
</property>
|
2021-09-27 16:57:04 +02:00
|
|
|
<property>
|
|
|
|
<name>openaireOrganizationPath</name>
|
|
|
|
<description>the OpenAire Organizations Path</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
|
2021-01-04 17:01:35 +01:00
|
|
|
<property>
|
|
|
|
<name>outputPath</name>
|
|
|
|
<description>the Path of the sequence file action set</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
|
|
|
|
<!-- Crossref Parameters -->
|
|
|
|
<property>
|
|
|
|
<name>inputPathCrossref</name>
|
|
|
|
<description>the Crossref input path</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<!-- MAG Parameters -->
|
|
|
|
<property>
|
|
|
|
<name>inputPathMAG</name>
|
2021-10-13 11:34:24 +02:00
|
|
|
<description>the MAG input path</description>
|
|
|
|
</property>
|
2021-01-04 17:01:35 +01:00
|
|
|
|
|
|
|
|
|
|
|
<!-- UnpayWall Parameters -->
|
|
|
|
<property>
|
|
|
|
<name>inputPathUnpayWall</name>
|
|
|
|
<description>the UnpayWall working path</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<!-- ORCID Parameters -->
|
2023-11-30 14:36:50 +01:00
|
|
|
<!-- <property>-->
|
|
|
|
<!-- <name>workingPathOrcid</name>-->
|
|
|
|
<!-- <description>the ORCID working path</description>-->
|
|
|
|
<!-- </property>-->
|
2021-01-04 17:01:35 +01:00
|
|
|
|
|
|
|
</parameters>
|
|
|
|
|
2021-01-05 10:00:13 +01:00
|
|
|
<global>
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<configuration>
|
|
|
|
<property>
|
|
|
|
<name>oozie.action.sharelib.for.spark</name>
|
|
|
|
<value>${oozieActionShareLibForSpark2}</value>
|
|
|
|
</property>
|
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2021-01-04 17:01:35 +01:00
|
|
|
<start to="resume_from"/>
|
|
|
|
|
|
|
|
<decision name="resume_from">
|
|
|
|
<switch>
|
2021-07-16 15:01:25 +02:00
|
|
|
<case to="End">${wf:conf('resumeFrom') eq 'Skip'}</case>
|
2021-01-05 10:26:39 +01:00
|
|
|
<case to="ProcessMAG">${wf:conf('resumeFrom') eq 'PreprocessMag'}</case>
|
|
|
|
<case to="ProcessUW">${wf:conf('resumeFrom') eq 'PreprocessUW'}</case>
|
2021-01-04 17:01:35 +01:00
|
|
|
<case to="CreateDOIBoost">${wf:conf('resumeFrom') eq 'CreateDOIBoost'}</case>
|
|
|
|
<case to="GenerateActionSet">${wf:conf('resumeFrom') eq 'GenerateActionSet'}</case>
|
2021-07-14 14:37:06 +02:00
|
|
|
<default to="ConvertCrossrefToOAF"/>
|
2021-01-04 17:01:35 +01:00
|
|
|
</switch>
|
|
|
|
</decision>
|
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2021-03-17 12:12:56 +01:00
|
|
|
<action name="ConvertCrossrefToOAF">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ConvertCrossrefToOAF</name>
|
|
|
|
<class>eu.dnetlib.doiboost.crossref.SparkMapDumpIntoOAF</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
2021-07-19 12:41:14 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
2021-03-17 12:12:56 +01:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${inputPathCrossref}/crossref_ds</arg>
|
|
|
|
<arg>--targetPath</arg><arg>${workingPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
2021-01-04 17:37:08 +01:00
|
|
|
<ok to="ProcessMAG"/>
|
2021-01-04 17:01:35 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2021-01-04 17:37:08 +01:00
|
|
|
<action name="ProcessMAG">
|
2021-01-04 17:01:35 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Convert Mag to OAF Dataset</name>
|
2021-01-04 17:37:08 +01:00
|
|
|
<class>eu.dnetlib.doiboost.mag.SparkProcessMAG</class>
|
2021-01-04 17:01:35 +01:00
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
2021-03-17 12:12:56 +01:00
|
|
|
--executor-memory=${sparkExecutorIntersectionMemory}
|
2021-01-04 17:01:35 +01:00
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
2021-07-19 18:16:23 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
2021-03-17 12:12:56 +01:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
2021-01-04 17:01:35 +01:00
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${inputPathMAG}/dataset</arg>
|
2021-10-13 15:33:50 +02:00
|
|
|
<arg>--workingPath</arg><arg>${workingPath}/MAG</arg>
|
2021-01-04 17:01:35 +01:00
|
|
|
<arg>--targetPath</arg><arg>${workingPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
2021-01-05 10:26:39 +01:00
|
|
|
<ok to="ProcessUW"/>
|
2021-01-04 17:01:35 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<!-- UnpayWall SECTION -->
|
|
|
|
|
2021-01-05 10:26:39 +01:00
|
|
|
<action name="ProcessUW">
|
2021-01-04 17:01:35 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Convert UnpayWall to Dataset</name>
|
|
|
|
<class>eu.dnetlib.doiboost.uw.SparkMapUnpayWallToOAF</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
2021-03-17 12:12:56 +01:00
|
|
|
--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}${spark2EventLogDir}
|
2021-01-04 17:01:35 +01:00
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${inputPathUnpayWall}/uw_extracted</arg>
|
2021-03-17 12:12:56 +01:00
|
|
|
<arg>--targetPath</arg><arg>${workingPath}/uwPublication</arg>
|
2021-01-04 17:01:35 +01:00
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="CreateDOIBoost"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2021-01-04 17:37:08 +01:00
|
|
|
<!-- INTERSECTION SECTION-->
|
2021-01-04 17:01:35 +01:00
|
|
|
<action name="CreateDOIBoost">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Create DOIBoost Infospace</name>
|
|
|
|
<class>eu.dnetlib.doiboost.SparkGenerateDoiBoost</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
2021-01-05 10:26:39 +01:00
|
|
|
--executor-memory=${sparkExecutorIntersectionMemory}
|
2021-01-04 17:01:35 +01:00
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
2021-03-17 12:12:56 +01:00
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
2021-01-04 17:01:35 +01:00
|
|
|
</spark-opts>
|
|
|
|
<arg>--hostedByMapPath</arg><arg>${hostedByMapPath}</arg>
|
2021-09-27 16:57:04 +02:00
|
|
|
<arg>--openaireOrganizationPath</arg><arg>${openaireOrganizationPath}</arg>
|
2021-03-17 12:12:56 +01:00
|
|
|
<arg>--affiliationPath</arg><arg>${inputPathMAG}/dataset/Affiliations</arg>
|
|
|
|
<arg>--paperAffiliationPath</arg><arg>${inputPathMAG}/dataset/PaperAuthorAffiliations</arg>
|
2021-01-04 17:37:08 +01:00
|
|
|
<arg>--workingPath</arg><arg>${workingPath}</arg>
|
2021-01-04 17:01:35 +01:00
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="GenerateActionSet"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
<action name="GenerateActionSet">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Generate DOIBoost ActionSet</name>
|
|
|
|
<class>eu.dnetlib.doiboost.SparkGenerateDOIBoostActionSet</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
2021-01-05 10:26:39 +01:00
|
|
|
--executor-memory=${sparkExecutorMemory}
|
2021-01-04 17:01:35 +01:00
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
2021-03-17 12:12:56 +01:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
2021-01-04 17:01:35 +01:00
|
|
|
</spark-opts>
|
|
|
|
<arg>--dbPublicationPath</arg><arg>${workingPath}/doiBoostPublicationFiltered</arg>
|
|
|
|
<arg>--dbDatasetPath</arg><arg>${workingPath}/crossrefDataset</arg>
|
|
|
|
<arg>--crossRefRelation</arg><arg>${workingPath}/crossrefRelation</arg>
|
|
|
|
<arg>--dbaffiliationRelationPath</arg><arg>${workingPath}/doiBoostPublicationAffiliation</arg>
|
2021-01-04 17:39:25 +01:00
|
|
|
<arg>--dbOrganizationPath</arg><arg>${workingPath}/doiBoostOrganization</arg>
|
2021-01-04 17:01:35 +01:00
|
|
|
<arg>--targetPath</arg><arg>${workingPath}/actionDataSet</arg>
|
|
|
|
<arg>--sFilePath</arg><arg>${outputPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|