|
|
|
@ -166,9 +166,9 @@
|
|
|
|
|
<!-- MAG SECTION -->
|
|
|
|
|
<action name="ResetMagWorkingPath">
|
|
|
|
|
<fs>
|
|
|
|
|
<delete path='${inputPathMAG}/dataset'/>
|
|
|
|
|
<delete path='${inputPathMAG}/process'/>
|
|
|
|
|
<delete path='${inputPathMAG}/dataset'/>
|
|
|
|
|
<delete path="${inputPathMAG}/dataset"/>
|
|
|
|
|
<delete path="${inputPathMAG}/process"/>
|
|
|
|
|
<delete path="${inputPathMAG}/dataset"/>
|
|
|
|
|
</fs>
|
|
|
|
|
<ok to="ConvertMagToDataset"/>
|
|
|
|
|
<error to="Kill"/>
|
|
|
|
@ -191,18 +191,16 @@
|
|
|
|
|
<arg>--targetPath</arg><arg>${inputPathMAG}/dataset</arg>
|
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
|
</spark>
|
|
|
|
|
<ok to="PreprocessMag"/>
|
|
|
|
|
<ok to="ProcessMAG"/>
|
|
|
|
|
<error to="Kill"/>
|
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<action name="PreprocessMag">
|
|
|
|
|
<action name="ProcessMAG">
|
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
|
<mode>cluster</mode>
|
|
|
|
|
<name>Convert Mag to OAF Dataset</name>
|
|
|
|
|
<class>eu.dnetlib.doiboost.mag.SparkPreProcessMAG</class>
|
|
|
|
|
<class>eu.dnetlib.doiboost.mag.SparkProcessMAG</class>
|
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
|
<spark-opts>
|
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
@ -244,8 +242,6 @@
|
|
|
|
|
<error to="Kill"/>
|
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<!-- ORCID SECTION -->
|
|
|
|
|
<action name="PreprocessORCID">
|
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
@ -269,7 +265,7 @@
|
|
|
|
|
<error to="Kill"/>
|
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
<!-- INTERSECTION SECTION-->
|
|
|
|
|
<!-- INTERSECTION SECTION-->
|
|
|
|
|
<action name="CreateDOIBoost">
|
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
@ -287,7 +283,7 @@
|
|
|
|
|
<arg>--hostedByMapPath</arg><arg>${hostedByMapPath}</arg>
|
|
|
|
|
<arg>--affiliationPath</arg><arg>${inputPathMAG}/process/Affiliations</arg>
|
|
|
|
|
<arg>--paperAffiliationPath</arg><arg>${inputPathMAG}/process/PaperAuthorAffiliations</arg>
|
|
|
|
|
<arg>--workingDirPath</arg><arg>${workingPath}</arg>
|
|
|
|
|
<arg>--workingPath</arg><arg>${workingPath}</arg>
|
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
|
</spark>
|
|
|
|
|
<ok to="GenerateActionSet"/>
|
|
|
|
|