orcid-no-doi #123

Merged
claudio.atzori merged 13 commits from enrico.ottonello/dnet-hadoop:orcid-no-doi into master 2021-07-15 17:53:59 +02:00
1 changed files with 7 additions and 4 deletions
Showing only changes of commit d0945c3c78 - Show all commits

View File

@ -9,7 +9,10 @@
<name>outputPath</name> <name>outputPath</name>
<description>path where to store the action set</description> <description>path where to store the action set</description>
</property> </property>
<property>
<name>processOutputFolder</name>
<description>temporary path where to store the action set</description>
</property>
<property> <property>
<name>spark2GenNoDoiDatasetMaxExecutors</name> <name>spark2GenNoDoiDatasetMaxExecutors</name>
<value>40</value> <value>40</value>
@ -66,7 +69,7 @@
<action name="ResetWorkingPath"> <action name="ResetWorkingPath">
<fs> <fs>
<delete path='${workingPath}/no_doi_dataset'/> <delete path='${workingPath}/${processOutputFolder}'/>
</fs> </fs>
<ok to="GenOrcidNoDoiDataset"/> <ok to="GenOrcidNoDoiDataset"/>
<error to="Kill"/> <error to="Kill"/>
@ -92,7 +95,7 @@
<arg>--workingPath</arg><arg>${workingPath}/</arg> <arg>--workingPath</arg><arg>${workingPath}/</arg>
<arg>--hdfsServerUri</arg><arg>${nameNode}</arg> <arg>--hdfsServerUri</arg><arg>${nameNode}</arg>
<arg>--orcidDataFolder</arg><arg>last_orcid_dataset</arg> <arg>--orcidDataFolder</arg><arg>last_orcid_dataset</arg>
<arg>--outputEnrichedWorksPath</arg><arg>no_doi_dataset</arg> <arg>--outputEnrichedWorksPath</arg><arg>${processOutputFolder}</arg>
</spark> </spark>
<ok to="importOrcidNoDoi"/> <ok to="importOrcidNoDoi"/>
<error to="Kill"/> <error to="Kill"/>
@ -100,7 +103,7 @@
<action name="importOrcidNoDoi"> <action name="importOrcidNoDoi">
<distcp xmlns="uri:oozie:distcp-action:0.2"> <distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${workingPath}/no_doi_dataset/*</arg> <arg>${workingPath}/${processOutputFolder}/*</arg>
<arg>${outputPath}</arg> <arg>${outputPath}</arg>
</distcp> </distcp>
<ok to="End"/> <ok to="End"/>