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

added global properties in wf definitions to avoid repeating name-node and job-tracker in the (many) distcp actions; reintroduced output directory removal at the beginning of each spark action

This commit is contained in:
Claudio Atzori 2020-05-14 10:25:41 +02:00
parent 12bfa6702e
commit ab37953332
17 changed files with 213 additions and 212 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/SparkBulkTagJob.java
View File

@ -1,6 +1,7 @@
package eu.dnetlib.dhp.bulktag;
import static eu.dnetlib.dhp.PropagationConstant.removeOutputDir;
import static eu.dnetlib.dhp.common.SparkSessionSupport.runWithSparkSession;
import java.util.Optional;
@ -84,6 +85,7 @@ public class SparkBulkTagJob {
conf,
isSparkSessionManaged,
spark -> {
removeOutputDir(spark, outputPath);
execBulkTag(spark, inputPath, outputPath, protoMappingParams, resultClazz, cc);
});
}

17
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/countrypropagation/SparkCountryPropagationJob.java
View File

@ -69,13 +69,16 @@ public class SparkCountryPropagationJob {
runWithSparkSession(
conf,
isSparkSessionManaged,
spark -> execPropagation(
spark,
sourcePath,
preparedInfoPath,
outputPath,
resultClazz,
saveGraph));
spark -> {
removeOutputDir(spark, outputPath);
execPropagation(
spark,
sourcePath,
preparedInfoPath,
outputPath,
resultClazz,
saveGraph);
});
}
private static <R extends Result> void execPropagation(

4
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/orcidtoresultfromsemrel/PrepareResultOrcidAssociationStep1.java
View File

@ -74,9 +74,7 @@ public class PrepareResultOrcidAssociationStep1 {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
removeOutputDir(spark, outputPath);
prepareInfo(
spark, inputRelationPath, inputResultPath, outputResultPath, resultClazz, allowedsemrel);
});

4
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/orcidtoresultfromsemrel/PrepareResultOrcidAssociationStep2.java
View File

@ -50,9 +50,7 @@ public class PrepareResultOrcidAssociationStep2 {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
removeOutputDir(spark, outputPath);
mergeInfo(spark, inputPath, outputPath);
});
}

7
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/orcidtoresultfromsemrel/SparkOrcidToResultFromSemRelJob.java
View File

@ -70,11 +70,10 @@ public class SparkOrcidToResultFromSemRelJob {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
if (saveGraph)
removeOutputDir(spark, outputPath);
if (saveGraph) {
execPropagation(spark, possibleUpdates, inputPath, outputPath, resultClazz);
}
});
}

2
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/projecttoresult/PrepareProjectResultsAssociation.java
View File

@ -60,6 +60,8 @@ public class PrepareProjectResultsAssociation {
conf,
isSparkSessionManaged,
spark -> {
removeOutputDir(spark, potentialUpdatePath);
removeOutputDir(spark, alreadyLinkedPath);
prepareResultProjProjectResults(
spark,
inputPath,

4
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/resulttocommunityfromorganization/PrepareResultCommunitySet.java
View File

@ -55,9 +55,7 @@ public class PrepareResultCommunitySet {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
removeOutputDir(spark, outputPath);
prepareInfo(spark, inputPath, outputPath, organizationMap);
});
}

7
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/resulttocommunityfromorganization/SparkResultToCommunityFromOrganizationJob.java
View File

@ -68,11 +68,10 @@ public class SparkResultToCommunityFromOrganizationJob {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
if (saveGraph)
removeOutputDir(spark, outputPath);
if (saveGraph) {
execPropagation(spark, inputPath, outputPath, resultClazz, possibleupdatespath);
}
});
}

43
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/PrepareResultInstRepoAssociation.java
View File

@ -58,30 +58,15 @@ public class PrepareResultInstRepoAssociation {
isSparkSessionManaged,
spark -> {
readNeededResources(spark, inputPath);
removeOutputDir(spark, datasourceOrganizationPath);
prepareDatasourceOrganization(spark, datasourceOrganizationPath);
removeOutputDir(spark, alreadyLinkedPath);
prepareAlreadyLinkedAssociation(spark, alreadyLinkedPath);
});
}
private static void prepareAlreadyLinkedAssociation(
SparkSession spark, String alreadyLinkedPath) {
String query = "Select source resultId, collect_set(target) organizationSet "
+ "from relation "
+ "where datainfo.deletedbyinference = false "
+ "and relClass = '"
+ RELATION_RESULT_ORGANIZATION_REL_CLASS
+ "' "
+ "group by source";
spark
.sql(query)
.as(Encoders.bean(ResultOrganizationSet.class))
// TODO retry to stick with datasets
.toJavaRDD()
.map(r -> OBJECT_MAPPER.writeValueAsString(r))
.saveAsTextFile(alreadyLinkedPath, GzipCodec.class);
}
private static void readNeededResources(SparkSession spark, String inputPath) {
Dataset<Datasource> datasource = readPath(spark, inputPath + "/datasource", Datasource.class);
datasource.createOrReplaceTempView("datasource");
@ -119,4 +104,24 @@ public class PrepareResultInstRepoAssociation {
.option("compression", "gzip")
.json(datasourceOrganizationPath);
}
private static void prepareAlreadyLinkedAssociation(
SparkSession spark, String alreadyLinkedPath) {
String query = "Select source resultId, collect_set(target) organizationSet "
+ "from relation "
+ "where datainfo.deletedbyinference = false "
+ "and relClass = '"
+ RELATION_RESULT_ORGANIZATION_REL_CLASS
+ "' "
+ "group by source";
spark
.sql(query)
.as(Encoders.bean(ResultOrganizationSet.class))
// TODO retry to stick with datasets
.toJavaRDD()
.map(r -> OBJECT_MAPPER.writeValueAsString(r))
.saveAsTextFile(alreadyLinkedPath, GzipCodec.class);
}
}

7
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/SparkResultToOrganizationFromIstRepoJob.java
View File

@ -83,10 +83,8 @@ public class SparkResultToOrganizationFromIstRepoJob {
conf,
isSparkSessionManaged,
spark -> {
if (isTest(parser)) {
removeOutputDir(spark, outputPath);
}
if (saveGraph)
removeOutputDir(spark, outputPath);
if (saveGraph) {
execPropagation(
spark,
datasourceorganization,
@ -94,6 +92,7 @@ public class SparkResultToOrganizationFromIstRepoJob {
inputPath,
outputPath,
resultClazz);
}
});
}

27
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/bulktag/oozie_app/workflow.xml
View File

@ -18,6 +18,17 @@
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="reset_outputpath"/>
<kill name="Kill">
@ -42,8 +53,6 @@
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
@ -53,8 +62,6 @@
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
@ -64,8 +71,6 @@
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
@ -75,8 +80,6 @@
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
@ -95,8 +98,6 @@
<action name="join_bulktag_publication">
<spark xmlns="uri:oozie:spark-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>bulkTagging-publication</name>
@ -124,8 +125,6 @@
<action name="join_bulktag_dataset">
<spark xmlns="uri:oozie:spark-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>bulkTagging-dataset</name>
@ -153,8 +152,6 @@
<action name="join_bulktag_otherresearchproduct">
<spark xmlns="uri:oozie:spark-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>bulkTagging-orp</name>
@ -182,8 +179,6 @@
<action name="join_bulktag_software">
<spark xmlns="uri:oozie:spark-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>bulkTagging-software</name>

20
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/countrypropagation/oozie_app/workflow.xml
View File

@ -19,6 +19,17 @@
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="reset_outputpath"/>
<kill name="Kill">
@ -43,8 +54,6 @@
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
@ -54,18 +63,15 @@
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
@ -75,8 +81,6 @@
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>

7
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/orcidtoresultfromsemrel/oozie_app/workflow.xml
View File

@ -57,6 +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>
@ -81,7 +82,6 @@
<join name="copy_wait" to="fork_prepare_assoc_step1"/>
<fork name="fork_prepare_assoc_step1">
<path start="join_prepare_publication"/>
<path start="join_prepare_dataset"/>
@ -230,8 +230,8 @@
</spark>
<ok to="fork-join-exec-propagation"/>
<error to="Kill"/>
</action>
<fork name="fork-join-exec-propagation">
<path start="join_propagate_publication"/>
<path start="join_propagate_dataset"/>
@ -271,6 +271,7 @@
<ok to="wait2"/>
<error to="Kill"/>
</action>
<action name="join_propagate_dataset">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
@ -302,6 +303,7 @@
<ok to="wait2"/>
<error to="Kill"/>
</action>
<action name="join_propagate_otherresearchproduct">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
@ -333,6 +335,7 @@
<ok to="wait2"/>
<error to="Kill"/>
</action>
<action name="join_propagate_software">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>

29
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/projecttoresult/oozie_app/workflow.xml
View File

@ -14,6 +14,17 @@
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="reset_outputpath"/>
<kill name="Kill">
@ -42,8 +53,6 @@
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
@ -53,8 +62,6 @@
<action name="copy_publication">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/publication</arg>
<arg>${nameNode}/${outputPath}/publication</arg>
</distcp>
@ -64,8 +71,6 @@
<action name="copy_dataset">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/dataset</arg>
<arg>${nameNode}/${outputPath}/dataset</arg>
</distcp>
@ -75,8 +80,6 @@
<action name="copy_orp">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/otherresearchproduct</arg>
<arg>${nameNode}/${outputPath}/otherresearchproduct</arg>
</distcp>
@ -86,28 +89,24 @@
<action name="copy_software">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/software</arg>
<arg>${nameNode}/${outputPath}/software</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
@ -117,8 +116,6 @@
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>

29
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttocommunityfromorganization/oozie_app/workflow.xml
View File

@ -14,6 +14,17 @@
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="reset_outputpath"/>
<kill name="Kill">
@ -38,8 +49,6 @@
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
@ -49,8 +58,6 @@
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
@ -60,8 +67,6 @@
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
@ -71,8 +76,6 @@
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
@ -101,8 +104,8 @@
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/relation</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--organizationtoresultcommunitymap</arg><arg>${organizationtoresultcommunitymap}</arg>
</spark>
<ok to="fork-join-exec-propagation"/>
@ -136,9 +139,9 @@
</spark-opts>
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
<arg>--outputPath</arg><arg>${outputPath}/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>${saveGraph}</arg>
</spark>
<ok to="wait2"/>
@ -165,9 +168,9 @@
</spark-opts>
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
<arg>--outputPath</arg><arg>${outputPath}/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>${saveGraph}</arg>
</spark>
<ok to="wait2"/>
@ -194,9 +197,9 @@
</spark-opts>
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
<arg>--outputPath</arg><arg>${outputPath}/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>${saveGraph}</arg>
</spark>
<ok to="wait2"/>
@ -223,9 +226,9 @@
</spark-opts>
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
<arg>--outputPath</arg><arg>${outputPath}/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"/>

52
dhp-workflows/dhp-enrichment/src/main/resources/eu/dnetlib/dhp/resulttoorganizationfrominstrepo/oozie_app/workflow.xml
View File

@ -10,6 +10,17 @@
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="reset_outputpath"/>
<kill name="Kill">
@ -38,8 +49,6 @@
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
@ -49,8 +58,6 @@
<action name="copy_publication">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/publication</arg>
<arg>${nameNode}/${outputPath}/publication</arg>
</distcp>
@ -60,8 +67,6 @@
<action name="copy_dataset">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/dataset</arg>
<arg>${nameNode}/${outputPath}/dataset</arg>
</distcp>
@ -71,8 +76,6 @@
<action name="copy_orp">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/otherresearchproduct</arg>
<arg>${nameNode}/${outputPath}/otherresearchproduct</arg>
</distcp>
@ -82,8 +85,6 @@
<action name="copy_software">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/software</arg>
<arg>${nameNode}/${outputPath}/software</arg>
</distcp>
@ -93,8 +94,6 @@
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
@ -104,8 +103,6 @@
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
@ -115,8 +112,6 @@
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
@ -125,6 +120,7 @@
</action>
<join name="wait" to="prepare_result_organization_association"/>
<action name="prepare_result_organization_association">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
@ -176,12 +172,12 @@
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputPath</arg><arg>${outputPath}/relation</arg>
<arg>--datasourceOrganizationPath</arg><arg>${workingDir}/preparedInfo/datasourceOrganization</arg>
<arg>--alreadyLinkedPath</arg><arg>${workingDir}/preparedInfo/alreadyLinked</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -206,12 +202,12 @@
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--outputPath</arg><arg>${outputPath}/relation</arg>
<arg>--datasourceOrganizationPath</arg><arg>${workingDir}/preparedInfo/datasourceOrganization</arg>
<arg>--alreadyLinkedPath</arg><arg>${workingDir}/preparedInfo/alreadyLinked</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -236,12 +232,12 @@
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--outputPath</arg><arg>${outputPath}/relation</arg>
<arg>--datasourceOrganizationPath</arg><arg>${workingDir}/preparedInfo/datasourceOrganization</arg>
<arg>--alreadyLinkedPath</arg><arg>${workingDir}/preparedInfo/alreadyLinked</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>
@ -266,12 +262,12 @@
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
</spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--outputPath</arg><arg>${outputPath}/relation</arg>
<arg>--datasourceOrganizationPath</arg><arg>${workingDir}/preparedInfo/datasourceOrganization</arg>
<arg>--alreadyLinkedPath</arg><arg>${workingDir}/preparedInfo/alreadyLinked</arg>
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
</spark>
<ok to="wait2"/>
<error to="Kill"/>

164
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/AbstractMdRecordToOafMapper.java
View File

@ -64,7 +64,7 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected static final Qualifier MAIN_TITLE_QUALIFIER = qualifier(
"main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
"main title", "main title", "dnet:dataCite_title", "dnet:dataCite_title");
protected AbstractMdRecordToOafMapper(final Map<String, String> code2name) {
this.code2name = code2name;
@ -75,20 +75,20 @@ public abstract class AbstractMdRecordToOafMapper {
DocumentFactory.getInstance().setXPathNamespaceURIs(nsContext);
final Document doc = DocumentHelper
.parseText(
xml.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3));
.parseText(
xml.replaceAll(DATACITE_SCHEMA_KERNEL_4, DATACITE_SCHEMA_KERNEL_3));
final String type = doc.valueOf("//dr:CobjCategory/@type");
final KeyValue collectedFrom = getProvenanceDatasource(
doc, "//oaf:collectedFrom/@id", "//oaf:collectedFrom/@name");
doc, "//oaf:collectedFrom/@id", "//oaf:collectedFrom/@name");
if (collectedFrom == null) {
return null;
}
final KeyValue hostedBy = StringUtils.isBlank(doc.valueOf("//oaf:hostedBy/@id"))
? collectedFrom
: getProvenanceDatasource(doc, "//oaf:hostedBy/@id", "//oaf:hostedBy/@name");
? collectedFrom
: getProvenanceDatasource(doc, "//oaf:hostedBy/@id", "//oaf:hostedBy/@name");
if (hostedBy == null) {
return null;
@ -112,17 +112,17 @@ public abstract class AbstractMdRecordToOafMapper {
}
return keyValue(
createOpenaireId(10, dsId, true),
dsName);
createOpenaireId(10, dsId, true),
dsName);
}
protected List<Oaf> createOafs(
final Document doc,
final String type,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
final Document doc,
final String type,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> oafs = new ArrayList<>();
@ -179,10 +179,10 @@ public abstract class AbstractMdRecordToOafMapper {
}
private List<Oaf> addProjectRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp) {
final List<Oaf> res = new ArrayList<>();
@ -196,15 +196,15 @@ public abstract class AbstractMdRecordToOafMapper {
final String projectId = createOpenaireId(40, originalId, true);
res
.add(
getRelation(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info,
lastUpdateTimestamp));
.add(
getRelation(
docId, projectId, RESULT_PROJECT, OUTCOME, IS_PRODUCED_BY, collectedFrom, info,
lastUpdateTimestamp));
res
.add(
getRelation(
projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info,
lastUpdateTimestamp));
.add(
getRelation(
projectId, docId, RESULT_PROJECT, OUTCOME, PRODUCES, collectedFrom, info,
lastUpdateTimestamp));
}
}
@ -212,7 +212,7 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected Relation getRelation(String source, String target, String relType, String subRelType, String relClass,
KeyValue collectedFrom, DataInfo info, long lastUpdateTimestamp) {
KeyValue collectedFrom, DataInfo info, long lastUpdateTimestamp) {
final Relation rel = new Relation();
rel.setRelType(relType);
rel.setSubRelType(subRelType);
@ -226,27 +226,27 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected abstract List<Oaf> addOtherResultRels(
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp);
final Document doc,
final KeyValue collectedFrom,
final DataInfo info,
final long lastUpdateTimestamp);
private void populateResultFields(
final Result r,
final Document doc,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
final Result r,
final Document doc,
final KeyValue collectedFrom,
final KeyValue hostedBy,
final DataInfo info,
final long lastUpdateTimestamp) {
r.setDataInfo(info);
r.setLastupdatetimestamp(lastUpdateTimestamp);
r.setId(createOpenaireId(50, doc.valueOf("//dri:objIdentifier"), false));
r.setOriginalId(Arrays.asList(doc.valueOf("//dri:objIdentifier")));
r.setCollectedfrom(Arrays.asList(collectedFrom));
r
.setPid(
prepareListStructProps(
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
.setPid(
prepareListStructProps(
doc, "//oaf:identifier", "@identifierType", "dnet:pid_types", "dnet:pid_types", info));
r.setDateofcollection(doc.valueOf("//dr:dateOfCollection"));
r.setDateoftransformation(doc.valueOf("//dr:dateOfTransformation"));
r.setExtraInfo(new ArrayList<>()); // NOT PRESENT IN MDSTORES
@ -289,7 +289,7 @@ public abstract class AbstractMdRecordToOafMapper {
protected abstract Qualifier prepareResourceType(Document doc, DataInfo info);
protected abstract List<Instance> prepareInstances(
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
Document doc, DataInfo info, KeyValue collectedfrom, KeyValue hostedby);
protected abstract List<Field<String>> prepareSources(Document doc, DataInfo info);
@ -314,13 +314,13 @@ public abstract class AbstractMdRecordToOafMapper {
protected abstract List<Author> prepareAuthors(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductTools(
Document doc, DataInfo info);
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactGroups(
Document doc, DataInfo info);
Document doc, DataInfo info);
protected abstract List<Field<String>> prepareOtherResearchProductContactPersons(
Document doc, DataInfo info);
Document doc, DataInfo info);
protected abstract Qualifier prepareSoftwareProgrammingLanguage(Document doc, DataInfo info);
@ -329,7 +329,7 @@ public abstract class AbstractMdRecordToOafMapper {
protected abstract List<StructuredProperty> prepareSoftwareLicenses(Document doc, DataInfo info);
protected abstract List<Field<String>> prepareSoftwareDocumentationUrls(
Document doc, DataInfo info);
Document doc, DataInfo info);
protected abstract List<GeoLocation> prepareDatasetGeoLocations(Document doc, DataInfo info);
@ -359,37 +359,37 @@ public abstract class AbstractMdRecordToOafMapper {
final String edition = n.valueOf("@edition");
if (StringUtils.isNotBlank(name)) {
return journal(
name,
issnPrinted,
issnOnline,
issnLinking,
ep,
iss,
sp,
vol,
edition,
null,
null,
info);
name,
issnPrinted,
issnOnline,
issnLinking,
ep,
iss,
sp,
vol,
edition,
null,
null,
info);
}
}
return null;
}
protected Qualifier prepareQualifier(
final Node node, final String xpath, final String schemeId, final String schemeName) {
final Node node, final String xpath, final String schemeId, final String schemeName) {
final String classId = node.valueOf(xpath);
final String className = code2name.get(classId);
return qualifier(classId, className, schemeId, schemeName);
}
protected List<StructuredProperty> prepareListStructProps(
final Node node,
final String xpath,
final String xpathClassId,
final String schemeId,
final String schemeName,
final DataInfo info) {
final Node node,
final String xpath,
final String xpathClassId,
final String schemeId,
final String schemeName,
final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
@ -401,7 +401,7 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
final Node node, final String xpath, final Qualifier qualifier, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
@ -411,19 +411,19 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected List<StructuredProperty> prepareListStructProps(
final Node node, final String xpath, final DataInfo info) {
final Node node, final String xpath, final DataInfo info) {
final List<StructuredProperty> res = new ArrayList<>();
for (final Object o : node.selectNodes(xpath)) {
final Node n = (Node) o;
res
.add(
structuredProperty(
n.getText(),
n.valueOf("@classid"),
n.valueOf("@classname"),
n.valueOf("@schemeid"),
n.valueOf("@schemename"),
info));
.add(
structuredProperty(
n.getText(),
n.valueOf("@classid"),
n.valueOf("@classname"),
n.valueOf("@schemeid"),
n.valueOf("@schemename"),
info));
}
return res;
}
@ -450,7 +450,7 @@ public abstract class AbstractMdRecordToOafMapper {
if (n == null) {
return dataInfo(
false, null, false, false, REPOSITORY_PROVENANCE_ACTIONS, "0.9");
false, null, false, false, REPOSITORY_PROVENANCE_ACTIONS, "0.9");
}
final String paClassId = n.valueOf("./oaf:provenanceaction/@classid");
@ -464,12 +464,12 @@ public abstract class AbstractMdRecordToOafMapper {
final String trust = n.valueOf("./oaf:trust");
return dataInfo(
deletedbyinference,
inferenceprovenance,
inferred,
false,
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
trust);
deletedbyinference,
inferenceprovenance,
inferred,
false,
qualifier(paClassId, paClassName, paSchemeId, paSchemeName),
trust);
}
protected Field<String> prepareField(final Node node, final String xpath, final DataInfo info) {
@ -477,7 +477,7 @@ public abstract class AbstractMdRecordToOafMapper {
}
protected List<Field<String>> prepareListFields(
final Node node, final String xpath, final DataInfo info) {
final Node node, final String xpath, final DataInfo info) {
return listFields(info, prepareListString(node, xpath));
}