sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

[dhp-actionmanager] missing descriptions added and minor naming and formatting fixes

This commit is contained in:
Przemysław Jacewicz 2020-04-02 11:48:40 +02:00 committed by przemek
parent 80cf43b9c8
commit 7b2a7e2417
14 changed files with 133 additions and 131 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/partition/PartitionActionSetsByPayloadTypeJob.java
View File

@ -73,7 +73,8 @@ public class PartitionActionSetsByPayloadTypeJob {
});
}
private static void removeOutputDir(SparkSession spark, String path) {
private static void removeOutputDir(SparkSession spark,
String path) {
HdfsSupport.remove(path, spark.sparkContext().hadoopConfiguration());
}

3
dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadForGraphTableJob.java
View File

@ -120,7 +120,8 @@ public class PromoteActionPayloadForGraphTableJob {
}
}
private static void removeOutputDir(SparkSession spark, String path) {
private static void removeOutputDir(SparkSession spark,
String path) {
HdfsSupport.remove(path, spark.sparkContext().hadoopConfiguration());
}

6
dhp-workflows/dhp-actionmanager/src/main/java/eu/dnetlib/dhp/actionmanager/promote/PromoteActionPayloadFunctions.java
View File

@ -37,8 +37,8 @@ public class PromoteActionPayloadFunctions {
* @param rowClazz Class of graph table
* @param actionPayloadClazz Class of action payload
* @param <G> Type of graph table row
* @param <A> Type of action payload
* @return Dataset of merged graph table rows and action payload
* @param <A> Type of action payload instance
* @return Dataset of merged graph table rows and action payload instances
*/
public static <G extends Oaf, A extends Oaf> Dataset<G> joinGraphTableWithActionPayloadAndMerge(Dataset<G> rowDS,
Dataset<A> actionPayloadDS,
@ -84,7 +84,7 @@ public class PromoteActionPayloadFunctions {
*
* @param rowDS Dataset representing graph table
* @param rowIdFn Function used to get the id of graph table row
* @param mergeAndGetFn Function used to merge graph table row and action payload instance
* @param mergeAndGetFn Function used to merge graph table rows
* @param zeroFn Function to create a zero/empty instance of graph table row
* @param isNotZeroFn Function to check if graph table row is not zero/empty
* @param rowClazz Class of graph table

8
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/partition/partition_action_sets_by_payload_type_input_parameters.json
View File

@ -2,19 +2,19 @@
{
"paramName": "issm",
"paramLongName": "isSparkSessionManaged",
"paramDescription": "TODO",
"paramDescription": "when true will stop SparkSession after job execution",
"paramRequired": false
},
{
"paramName": "iasp",
"paramLongName": "inputActionSetPaths",
"paramDescription": "TODO",
"paramDescription": "comma separated list of action sets to partition by payload type",
"paramRequired": true
},
{
"paramName": "opasp",
"paramName": "op",
"paramLongName": "outputPath",
"paramDescription": "TODO",
"paramDescription": "root output location for partitioned action sets",
"paramRequired": true
}
]

14
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/promote/promote_action_payload_for_graph_table_input_parameters.json
View File

@ -2,43 +2,43 @@
{
"paramName": "issm",
"paramLongName": "isSparkSessionManaged",
"paramDescription": "TODO",
"paramDescription": "when true will stop SparkSession after job execution",
"paramRequired": false
},
{
"paramName": "igtp",
"paramLongName": "inputGraphTablePath",
"paramDescription": "TODO",
"paramDescription": "location of graph table to promote",
"paramRequired": true
},
{
"paramName": "gtcn",
"paramLongName": "graphTableClassName",
"paramDescription": "TODO",
"paramDescription": "canonical name of graph table row class",
"paramRequired": true
},
{
"paramName": "iapp",
"paramLongName": "inputActionPayloadPath",
"paramDescription": "TODO",
"paramDescription": "location of action payload to promote",
"paramRequired": true
},
{
"paramName": "apcn",
"paramLongName": "actionPayloadClassName",
"paramDescription": "TODO",
"paramDescription": "canonical name of action payload class",
"paramRequired": true
},
{
"paramName": "ogtp",
"paramLongName": "outputGraphTablePath",
"paramDescription": "TODO",
"paramDescription": "location of promoted graph table",
"paramRequired": true
},
{
"paramName": "mags",
"paramLongName": "mergeAndGetStrategy",
"paramDescription": "TODO",
"paramDescription": "strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET",
"paramRequired": true
}
]

26
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml
View File

@ -2,27 +2,27 @@
<parameters>
<property>
<name>activePromoteDatasetActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Dataset payload</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Result payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -81,7 +81,7 @@
<switch>
<case to="PromoteDatasetActionPayloadForDatasetTable">
${(activePromoteDatasetActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'dataset')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'dataset')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Dataset')) eq "true")}
</case>
<default to="SkipPromoteDatasetActionPayloadForDatasetTable"/>
@ -105,7 +105,7 @@
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/dataset</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/dataset</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
@ -119,7 +119,7 @@
<action name="SkipPromoteDatasetActionPayloadForDatasetTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/dataset</arg>
<arg>${inputGraphRootPath}/dataset</arg>
<arg>${workingDir}/dataset</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForDatasetTable"/>
@ -157,7 +157,7 @@
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/dataset</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/dataset</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -168,7 +168,7 @@
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/dataset</arg>
<arg>${outputGraphPath}/dataset</arg>
<arg>${outputGraphRootPath}/dataset</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

24
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml
View File

@ -2,23 +2,23 @@
<parameters>
<property>
<name>activePromoteDatasourceActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Datasource payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -77,7 +77,7 @@
<switch>
<case to="PromoteDatasourceActionPayloadForDatasourceTable">
${(activePromoteDatasourceActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'datasource')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'datasource')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Datasource')) eq "true")}
</case>
<default to="SkipPromoteDatasourceActionPayloadForDatasourceTable"/>
@ -100,11 +100,11 @@
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/datasource</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/datasource</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/datasource</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/datasource</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -114,8 +114,8 @@
<action name="SkipPromoteDatasourceActionPayloadForDatasourceTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/datasource</arg>
<arg>${outputGraphPath}/datasource</arg>
<arg>${inputGraphRootPath}/datasource</arg>
<arg>${outputGraphRootPath}/datasource</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

30
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/main/oozie_app/workflow.xml
View File

@ -2,55 +2,55 @@
<parameters>
<property>
<name>activePromoteDatasetActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Dataset payload</description>
</property>
<property>
<name>activePromoteDatasourceActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Datasource payload</description>
</property>
<property>
<name>activePromoteOrganizationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Organization payload</description>
</property>
<property>
<name>activePromoteOtherResearchProductActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.OtherResearchProduct payload</description>
</property>
<property>
<name>activePromoteProjectActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Project payload</description>
</property>
<property>
<name>activePromotePublicationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Publication payload</description>
</property>
<property>
<name>activePromoteRelationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Relation payload</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Result payload</description>
</property>
<property>
<name>activePromoteSoftwareActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Software payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionSetPaths</name>
<description>TODO</description>
<description>comma separated list of action sets to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>

24
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml
View File

@ -2,23 +2,23 @@
<parameters>
<property>
<name>activePromoteOrganizationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Organization payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -77,7 +77,7 @@
<switch>
<case to="PromoteOrganizationActionPayloadForOrganizationTable">
${(activePromoteOrganizationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'organization')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'organization')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Organization')) eq "true")}
</case>
<default to="SkipPromoteOrganizationActionPayloadForOrganizationTable"/>
@ -100,11 +100,11 @@
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/organization</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/organization</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/organization</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/organization</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -114,8 +114,8 @@
<action name="SkipPromoteOrganizationActionPayloadForOrganizationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/organization</arg>
<arg>${outputGraphPath}/organization</arg>
<arg>${inputGraphRootPath}/organization</arg>
<arg>${outputGraphRootPath}/organization</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

26
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml
View File

@ -2,27 +2,27 @@
<parameters>
<property>
<name>activePromoteOtherResearchProductActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.OtherResearchProduct payload</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Result payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -81,7 +81,7 @@
<switch>
<case to="PromoteOtherResearchProductActionPayloadForOtherResearchProductTable">
${(activePromoteOtherResearchProductActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'otherresearchproduct')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'otherresearchproduct')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.OtherResearchProduct')) eq "true")}
</case>
<default to="SkipPromoteOtherResearchProductActionPayloadForOtherResearchProductTable"/>
@ -104,7 +104,7 @@
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/otherresearchproduct</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/otherresearchproduct</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
@ -118,7 +118,7 @@
<action name="SkipPromoteOtherResearchProductActionPayloadForOtherResearchProductTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/otherresearchproduct</arg>
<arg>${inputGraphRootPath}/otherresearchproduct</arg>
<arg>${workingDir}/otherresearchproduct</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForOtherResearchProductTable"/>
@ -156,7 +156,7 @@
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/otherresearchproduct</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/otherresearchproduct</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -167,7 +167,7 @@
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/otherresearchproduct</arg>
<arg>${outputGraphPath}/otherresearchproduct</arg>
<arg>${outputGraphRootPath}/otherresearchproduct</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

24
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml
View File

@ -2,23 +2,23 @@
<parameters>
<property>
<name>activePromoteProjectActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.OtherResearchProduct payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -77,7 +77,7 @@
<switch>
<case to="PromoteProjectActionPayloadForProjectTable">
${(activePromoteProjectActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'project')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'project')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Project')) eq "true")}
</case>
<default to="SkipPromoteProjectActionPayloadForProjectTable"/>
@ -100,11 +100,11 @@
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/project</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/project</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/project</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/project</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -114,8 +114,8 @@
<action name="SkipPromoteProjectActionPayloadForProjectTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/project</arg>
<arg>${outputGraphPath}/project</arg>
<arg>${inputGraphRootPath}/project</arg>
<arg>${outputGraphRootPath}/project</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

26
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml
View File

@ -2,27 +2,27 @@
<parameters>
<property>
<name>activePromotePublicationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Publication payload</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Result payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -81,7 +81,7 @@
<switch>
<case to="PromotePublicationActionPayloadForPublicationTable">
${(activePromotePublicationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'publication')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'publication')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Publication')) eq "true")}
</case>
<default to="SkipPromotePublicationActionPayloadForPublicationTable"/>
@ -105,7 +105,7 @@
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/publication</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/publication</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
@ -119,7 +119,7 @@
<action name="SkipPromotePublicationActionPayloadForPublicationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/publication</arg>
<arg>${inputGraphRootPath}/publication</arg>
<arg>${workingDir}/publication</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForPublicationTable"/>
@ -157,7 +157,7 @@
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/publication</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/publication</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -168,7 +168,7 @@
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/publication</arg>
<arg>${outputGraphPath}/publication</arg>
<arg>${outputGraphRootPath}/publication</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

24
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml
View File

@ -2,23 +2,23 @@
<parameters>
<property>
<name>activePromoteRelationActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Relation payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -77,7 +77,7 @@
<switch>
<case to="PromoteRelationActionPayloadForRelationTable">
${(activePromoteRelationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'relation')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'relation')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Relation')) eq "true")}
</case>
<default to="SkipPromoteRelationActionPayloadForRelationTable"/>
@ -101,11 +101,11 @@
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/relation</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/relation</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/relation</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/relation</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -115,8 +115,8 @@
<action name="SkipPromoteRelationActionPayloadForRelationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/relation</arg>
<arg>${outputGraphPath}/relation</arg>
<arg>${inputGraphRootPath}/relation</arg>
<arg>${outputGraphRootPath}/relation</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>

26
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml
View File

@ -2,27 +2,27 @@
<parameters>
<property>
<name>activePromoteSoftwareActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Software payload</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
<description>when true will promote actions with eu.dnetlib.dhp.schema.oaf.Result payload</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
<name>inputGraphRootPath</name>
<description>root location of input materialized graph</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
<description>root location of action payloads to promote</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
<name>outputGraphRootPath</name>
<description>root location for output materialized graph</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</description>
<description>strategy for merging graph table objects with action payload instances, MERGE_FROM_AND_GET or SELECT_NEWER_AND_GET</description>
</property>
<property>
<name>sparkDriverMemory</name>
@ -81,7 +81,7 @@
<switch>
<case to="PromoteSoftwareActionPayloadForSoftwareTable">
${(activePromoteSoftwareActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'software')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphRootPath')),'/'),'software')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Software')) eq "true")}
</case>
<default to="SkipPromoteSoftwareActionPayloadForSoftwareTable"/>
@ -104,7 +104,7 @@
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/software</arg>
<arg>--inputGraphTablePath</arg><arg>${inputGraphRootPath}/software</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
@ -118,7 +118,7 @@
<action name="SkipPromoteSoftwareActionPayloadForSoftwareTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/software</arg>
<arg>${inputGraphRootPath}/software</arg>
<arg>${workingDir}/software</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForSoftwareTable"/>
@ -156,7 +156,7 @@
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/software</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphRootPath}/software</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
@ -167,7 +167,7 @@
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/software</arg>
<arg>${outputGraphPath}/software</arg>
<arg>${outputGraphRootPath}/software</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>