simplified reset_outputpath nodes across the workflows, applied common xml formatting

dhp-workflows/dhp-bulktag/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/workflow.xml

@@ -1,39 +1,33 @@
 <workflow-app name="bulk_tagging" xmlns="uri:oozie:workflow:0.5">
-<parameters>
-    <property>
-        <name>sourcePath</name>
-        <description>the source path</description>
-    </property>
-    <property>
-        <name>isLookUpUrl</name>
-        <description>the isLookup service endpoint</description>
-    </property>
-    <property>
-        <name>pathMap</name>
-        <description>the json path associated to each selection field</description>
-    </property>
-    <property>
-        <name>outputPath</name>
-        <description>the output path</description>
-    </property>
-</parameters>
+    <parameters>
+        <property>
+            <name>sourcePath</name>
+            <description>the source path</description>
+        </property>
+        <property>
+            <name>isLookUpUrl</name>
+            <description>the isLookup service endpoint</description>
+        </property>
+        <property>
+            <name>pathMap</name>
+            <description>the json path associated to each selection field</description>
+        </property>
+        <property>
+            <name>outputPath</name>
+            <description>the output path</description>
+        </property>
+    </parameters>
 
-<start to="reset-outputpath"/>
+    <start to="reset_outputpath"/>
 
-<kill name="Kill">
-    <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
-</kill>
+    <kill name="Kill">
+        <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
+    </kill>
 
-    <action name="reset-outputpath">
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/datasource'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
@@ -67,6 +61,7 @@
         <ok to="copy_wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="copy_projects">
         <distcp xmlns="uri:oozie:distcp-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -91,14 +86,13 @@
 
     <join name="copy_wait" to="fork_exec_bulktag"/>
 
+    <fork name="fork_exec_bulktag">
+        <path start="join_bulktag_publication"/>
+        <path start="join_bulktag_dataset"/>
+        <path start="join_bulktag_otherresearchproduct"/>
+        <path start="join_bulktag_software"/>
+    </fork>
 
-
-<fork name="fork_exec_bulktag">
-    <path start="join_bulktag_publication"/>
-    <path start="join_bulktag_dataset"/>
-    <path start="join_bulktag_otherresearchproduct"/>
-    <path start="join_bulktag_software"/>
-</fork>
     <action name="join_bulktag_publication">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -127,6 +121,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_bulktag_dataset">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -155,6 +150,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_bulktag_otherresearchproduct">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -183,6 +179,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_bulktag_software">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -211,6 +208,9 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
-<join name="wait" to="End"/>
-<end name="End"/>
+
+    <join name="wait" to="End"/>
+
+    <end name="End"/>
+
 </workflow-app>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/workflow.xml

@@ -27,14 +27,8 @@
 
     <action name="reset_outputpath">
         <fs>
-            <delete path="${outputPath}/relation"/>
-            <delete path="${outputPath}/dataset"/>
-            <delete path="${outputPath}/software"/>
-            <delete path="${outputPath}/publication"/>
-            <delete path="${outputPath}/otherresearchproduct"/>
-            <delete path="${outputPath}/project"/>
-            <delete path="${outputPath}/organization"/>
-            <delete path="${outputPath}/datasource"/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/workflow.xml

@@ -1,37 +1,29 @@
 <workflow-app name="orcid_to_result_from_semrel_propagation" xmlns="uri:oozie:workflow:0.5">
-<parameters>
-    <property>
-        <name>sourcePath</name>
-        <description>the source path</description>
-    </property>
-    <property>
-        <name>allowedsemrels</name>
-        <description>the semantic relationships allowed for propagation</description>
-    </property>
-    <property>
-        <name>outputPath</name>
-        <description>the output path</description>
-    </property>
-</parameters>
-    <start to="reset-outputpath"/>
+    <parameters>
+        <property>
+            <name>sourcePath</name>
+            <description>the source path</description>
+        </property>
+        <property>
+            <name>allowedsemrels</name>
+            <description>the semantic relationships allowed for propagation</description>
+        </property>
+        <property>
+            <name>outputPath</name>
+            <description>the output path</description>
+        </property>
+    </parameters>
 
+    <start to="reset_outputpath"/>
 
-<!--    <start to="fork_prepare_assoc_step1"/>-->
+    <kill name="Kill">
+        <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
+    </kill>
 
-<kill name="Kill">
-    <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
-</kill>
-    <action name="reset-outputpath">
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${workingDir}/preparedInfo'/>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/datasource'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
@@ -119,11 +111,16 @@
                 --conf spark.hadoop.mapreduce.map.speculative=false
                 --conf spark.hadoop.mapreduce.reduce.speculative=false
             </spark-opts>
-            <arg>--sourcePath</arg><arg>${sourcePath}</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
-            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
-            <arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
+            <arg>--outputPath</arg>
+            <arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
+            <arg>--allowedsemrels</arg>
+            <arg>${allowedsemrels}</arg>
         </spark>
         <ok to="wait"/>
         <error to="Kill"/>
@@ -147,11 +144,16 @@
                 --conf spark.dynamicAllocation.enabled=true
                 --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
             </spark-opts>
-            <arg>--sourcePath</arg><arg>${sourcePath}</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
-            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
-            <arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
+            <arg>--outputPath</arg>
+            <arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
+            <arg>--allowedsemrels</arg>
+            <arg>${allowedsemrels}</arg>
         </spark>
         <ok to="wait"/>
         <error to="Kill"/>
@@ -175,11 +177,16 @@
                 --conf spark.dynamicAllocation.enabled=true
                 --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
             </spark-opts>
-            <arg>--sourcePath</arg><arg>${sourcePath}</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
-            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
-            <arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
+            <arg>--outputPath</arg>
+            <arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
+            <arg>--allowedsemrels</arg>
+            <arg>${allowedsemrels}</arg>
         </spark>
         <ok to="wait"/>
         <error to="Kill"/>
@@ -203,11 +210,16 @@
                 --conf spark.dynamicAllocation.enabled=true
                 --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
             </spark-opts>
-            <arg>--sourcePath</arg><arg>${sourcePath}</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
-            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
-            <arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
+            <arg>--outputPath</arg>
+            <arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
+            <arg>--allowedsemrels</arg>
+            <arg>${allowedsemrels}</arg>
         </spark>
         <ok to="wait"/>
         <error to="Kill"/>
@@ -233,11 +245,13 @@
                 --conf spark.dynamicAllocation.enabled=true
                 --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
             </spark-opts>
-            <arg>--sourcePath</arg><arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
-            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
+            <arg>--sourcePath</arg>
+            <arg>${workingDir}/preparedInfo/targetOrcidAssoc</arg>
+            <arg>--outputPath</arg>
+            <arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
         </spark>
         <ok to="fork-join-exec-propagation"/>
-<!--        <ok to="End"/>-->
+        <!--        <ok to="End"/>-->
         <error to="Kill"/>
 
     </action>
@@ -270,12 +284,18 @@
                 --conf spark.hadoop.mapreduce.reduce.speculative=false
                 --conf spark.sql.shuffle.partitions=3840
             </spark-opts>
-            <arg>--possibleUpdatesPath</arg><arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
-            <arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
-            <arg>--outputPath</arg><arg>${outputPath}/publication</arg>
-            <arg>--saveGraph</arg><arg>${saveGraph}</arg>
+            <arg>--possibleUpdatesPath</arg>
+            <arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}/publication</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
+            <arg>--outputPath</arg>
+            <arg>${outputPath}/publication</arg>
+            <arg>--saveGraph</arg>
+            <arg>${saveGraph}</arg>
         </spark>
         <ok to="wait2"/>
         <error to="Kill"/>
@@ -301,12 +321,18 @@
                 --conf spark.hadoop.mapreduce.map.speculative=false
                 --conf spark.hadoop.mapreduce.reduce.speculative=false
             </spark-opts>
-            <arg>--possibleUpdatesPath</arg><arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
-            <arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
-            <arg>--outputPath</arg><arg>${outputPath}/dataset</arg>
-            <arg>--saveGraph</arg><arg>${saveGraph}</arg>
+            <arg>--possibleUpdatesPath</arg>
+            <arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}/dataset</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
+            <arg>--outputPath</arg>
+            <arg>${outputPath}/dataset</arg>
+            <arg>--saveGraph</arg>
+            <arg>${saveGraph}</arg>
         </spark>
         <ok to="wait2"/>
         <error to="Kill"/>
@@ -332,12 +358,18 @@
                 --conf spark.hadoop.mapreduce.map.speculative=false
                 --conf spark.hadoop.mapreduce.reduce.speculative=false
             </spark-opts>
-            <arg>--possibleUpdatesPath</arg><arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
-            <arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
-            <arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg>
-            <arg>--saveGraph</arg><arg>${saveGraph}</arg>
+            <arg>--possibleUpdatesPath</arg>
+            <arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}/otherresearchproduct</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
+            <arg>--outputPath</arg>
+            <arg>${outputPath}/otherresearchproduct</arg>
+            <arg>--saveGraph</arg>
+            <arg>${saveGraph}</arg>
         </spark>
         <ok to="wait2"/>
         <error to="Kill"/>
@@ -363,16 +395,22 @@
                 --conf spark.hadoop.mapreduce.map.speculative=false
                 --conf spark.hadoop.mapreduce.reduce.speculative=false
             </spark-opts>
-            <arg>--possibleUpdatesPath</arg><arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
-            <arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
-            <arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
-            <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
-            <arg>--outputPath</arg><arg>${outputPath}/software</arg>
-            <arg>--saveGraph</arg><arg>${saveGraph}</arg>
+            <arg>--possibleUpdatesPath</arg>
+            <arg>${workingDir}/preparedInfo/mergedOrcidAssoc</arg>
+            <arg>--sourcePath</arg>
+            <arg>${sourcePath}/software</arg>
+            <arg>--hive_metastore_uris</arg>
+            <arg>${hive_metastore_uris}</arg>
+            <arg>--resultTableName</arg>
+            <arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
+            <arg>--outputPath</arg>
+            <arg>${outputPath}/software</arg>
+            <arg>--saveGraph</arg>
+            <arg>${saveGraph}</arg>
         </spark>
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
     <join name="wait2" to="End"/>
-<end name="End"/>
+    <end name="End"/>
 </workflow-app>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/workflow.xml

@@ -14,27 +14,21 @@
        </property>
     </parameters>
 
-    <start to="reset-outputpath"/>
-<!--    <start to="apply_propagation"/>-->
+    <start to="reset_outputpath"/>
+
     <kill name="Kill">
         <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
     </kill>
 
-    <action name="reset-outputpath">
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${workingDir}/preparedInfo'/>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/datasource'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
     </action>
+
     <fork name="copy_entities">
         <path start="copy_relation"/>
         <path start="copy_publication"/>
@@ -190,4 +184,5 @@
         </action>
 
     <end name="End"/>
+
 </workflow-app>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/workflow.xml

@@ -14,23 +14,16 @@
         </property>
     </parameters>
 
-    <start to="reset-outputpath"/>
+    <start to="reset_outputpath"/>
 
     <kill name="Kill">
         <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
     </kill>
 
-    <action name="reset-outputpath">
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/datasource'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${workingDir}/preparedInfo'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
@@ -64,6 +57,7 @@
         <ok to="copy_wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="copy_projects">
         <distcp xmlns="uri:oozie:distcp-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -85,6 +79,7 @@
         <ok to="copy_wait"/>
         <error to="Kill"/>
     </action>
+
     <join name="copy_wait" to="prepare_result_communitylist"/>
 
     <action name="prepare_result_communitylist">
@@ -111,7 +106,6 @@
             <arg>--organizationtoresultcommunitymap</arg><arg>${organizationtoresultcommunitymap}</arg>
         </spark>
         <ok to="fork-join-exec-propagation"/>
-<!--        <ok to="End"/>-->
         <error to="Kill"/>
     </action>
 
@@ -150,6 +144,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_dataset">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -178,6 +173,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_otherresearchproduct">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -206,6 +202,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_software">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -238,4 +235,5 @@
     <join name="wait2" to="End"/>
 
     <end name="End"/>
+
 </workflow-app>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromsemrel/oozie_app/workflow.xml

@@ -1,43 +1,38 @@
 <workflow-app name="result_to_community_from_semrel_propagation" xmlns="uri:oozie:workflow:0.5">
-<parameters>
-    <property>
-        <name>sourcePath</name>
-        <description>the source path</description>
-    </property>
-    <property>
-        <name>allowedsemrels</name>
-        <description>the semantic relationships allowed for propagation</description>
-    </property>
-    <property>
-        <name>isLookUpUrl</name>
-        <description>the isLookup service endpoint</description>
-    </property>
-    <property>
-        <name>outputPath</name>
-        <description>the output path</description>
-    </property>
-</parameters>
+    <parameters>
+        <property>
+            <name>sourcePath</name>
+            <description>the source path</description>
+        </property>
+        <property>
+            <name>allowedsemrels</name>
+            <description>the semantic relationships allowed for propagation</description>
+        </property>
+        <property>
+            <name>isLookUpUrl</name>
+            <description>the isLookup service endpoint</description>
+        </property>
+        <property>
+            <name>outputPath</name>
+            <description>the output path</description>
+        </property>
+    </parameters>
 
-    <start to="reset-outputpath"/>
+    <start to="reset_outputpath"/>
 
     <kill name="Kill">
         <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
     </kill>
-    <action name="reset-outputpath">
+
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/datasource'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${workingDir}/preparedInfo'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
     </action>
+
     <fork name="copy_entities">
         <path start="copy_relation"/>
         <path start="copy_organization"/>
@@ -126,6 +121,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_prepare_dataset">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -182,6 +178,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_prepare_software">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -214,29 +211,29 @@
     <join name="wait" to="prepare_assoc_step2"/>
 
     <action name="prepare_assoc_step2">
-     <spark xmlns="uri:oozie:spark-action:0.2">
-        <master>yarn</master>
-        <mode>cluster</mode>
-        <name>ResultToCommunityEmRelPropagation-PreparePhase2</name>
-        <class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep2</class>
-        <jar>dhp-propagation-${projectVersion}.jar</jar>
-        <spark-opts>
-            --executor-cores=${sparkExecutorCores}
-            --executor-memory=${sparkExecutorMemory}
-            --driver-memory=${sparkDriverMemory}
-            --conf spark.extraListeners=${spark2ExtraListeners}
-            --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
-            --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
-            --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
-            --conf spark.dynamicAllocation.enabled=true
-            --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
-        </spark-opts>
-        <arg>--sourcePath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
-        <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
-    </spark>
-    <ok to="fork-join-exec-propagation"/>
-<!--            <ok to="End"/>-->
-    <error to="Kill"/>
+        <spark xmlns="uri:oozie:spark-action:0.2">
+            <master>yarn</master>
+            <mode>cluster</mode>
+            <name>ResultToCommunityEmRelPropagation-PreparePhase2</name>
+            <class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep2</class>
+            <jar>dhp-propagation-${projectVersion}.jar</jar>
+            <spark-opts>
+                --executor-cores=${sparkExecutorCores}
+                --executor-memory=${sparkExecutorMemory}
+                --driver-memory=${sparkDriverMemory}
+                --conf spark.extraListeners=${spark2ExtraListeners}
+                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
+                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
+                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
+                --conf spark.dynamicAllocation.enabled=true
+                --conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
+            </spark-opts>
+            <arg>--sourcePath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
+            <arg>--outputPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
+        </spark>
+        <ok to="fork-join-exec-propagation"/>
+        <!--            <ok to="End"/>-->
+        <error to="Kill"/>
 
     </action>
 
@@ -275,6 +272,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_dataset">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -303,6 +301,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_otherresearchproduct">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -331,6 +330,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagate_software">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -362,5 +362,6 @@
 
     <join name="wait2" to="End"/>
 
-<end name="End"/>
+    <end name="End"/>
+
 </workflow-app>

dhp-workflows/dhp-propagation/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/workflow.xml

@@ -1,32 +1,25 @@
 <workflow-app name="affiliation_from_instrepo_propagation" xmlns="uri:oozie:workflow:0.5">
-<parameters>
-    <property>
-        <name>sourcePath</name>
-        <description>the source path</description>
-    </property>
-    <property>
-        <name>outputPath</name>
-        <description>sets the outputPath</description>
-    </property>
-</parameters>
+    <parameters>
+        <property>
+            <name>sourcePath</name>
+            <description>the source path</description>
+        </property>
+        <property>
+            <name>outputPath</name>
+            <description>sets the outputPath</description>
+        </property>
+    </parameters>
 
-<start to="reset-outputpath"/>
-<!--<start to="prepare_result_organization_association"/>-->
-<kill name="Kill">
-    <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
-</kill>
+    <start to="reset_outputpath"/>
 
-    <action name="reset-outputpath">
+    <kill name="Kill">
+        <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
+    </kill>
+
+    <action name="reset_outputpath">
         <fs>
-            <delete path='${outputPath}/relation'/>
-            <delete path='${outputPath}/datasource'/>
-            <delete path='${outputPath}/organization'/>
-            <delete path='${outputPath}/project'/>
-            <delete path='${outputPath}/publication'/>
-            <delete path='${outputPath}/dataset'/>
-            <delete path='${outputPath}/software'/>
-            <delete path='${outputPath}/otherresearchproduct'/>
-            <delete path='${workingDir}/preparedInfo'/>
+            <delete path="${outputPath}"/>
+            <mkdir path="${outputPath}"/>
         </fs>
         <ok to="copy_entities"/>
         <error to="Kill"/>
@@ -53,6 +46,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="copy_publication">
         <distcp xmlns="uri:oozie:distcp-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -85,6 +79,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="copy_software">
         <distcp xmlns="uri:oozie:distcp-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -95,6 +90,7 @@
         <ok to="wait"/>
         <error to="Kill"/>
     </action>
+
     <action name="copy_organization">
         <distcp xmlns="uri:oozie:distcp-action:0.2">
             <job-tracker>${jobTracker}</job-tracker>
@@ -151,9 +147,9 @@
             <arg>--alreadyLinkedPath</arg><arg>${workingDir}/preparedInfo/alreadyLinked</arg>
         </spark>
         <ok to="fork_join_apply_resulttoorganization_propagation"/>
-        <!--        <ok to="End"/>-->
         <error to="Kill"/>
     </action>
+
     <fork name="fork_join_apply_resulttoorganization_propagation">
         <path start="join_propagation_publication"/>
         <path start="join_propagation_dataset"/>
@@ -220,6 +216,7 @@
         <ok to="wait2"/>
         <error to="Kill"/>
     </action>
+
     <action name="join_propagation_otherresearchproduct">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn</master>
@@ -281,6 +278,7 @@
     </action>
 
     <join name="wait2" to="End"/>
+
     <end name="End"/>
 
 </workflow-app>