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

code formatting

This commit is contained in:
Claudio Atzori 2024-04-20 08:10:58 +02:00
parent ab7f0855af
commit 0656ab2838
9 changed files with 80 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
dhp-pace-core/src/main/java/eu/dnetlib/pace/tree/AuthorsMatch.java
View File

@ -1,19 +1,20 @@
package eu.dnetlib.pace.tree;
import com.wcohen.ss.AbstractStringDistance;
import eu.dnetlib.pace.config.Config;
import eu.dnetlib.pace.model.Person;
import eu.dnetlib.pace.tree.support.AbstractListComparator;
import eu.dnetlib.pace.tree.support.ComparatorClass;
import eu.dnetlib.pace.util.AuthorMatchers;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.function.BiFunction;
import java.util.stream.Collectors;
import com.wcohen.ss.AbstractStringDistance;
import eu.dnetlib.pace.config.Config;
import eu.dnetlib.pace.model.Person;
import eu.dnetlib.pace.tree.support.AbstractListComparator;
import eu.dnetlib.pace.tree.support.ComparatorClass;
import eu.dnetlib.pace.util.AuthorMatchers;
@ComparatorClass("authorsMatch")
public class AuthorsMatch extends AbstractListComparator {

1
dhp-pace-core/src/test/java/eu/dnetlib/pace/util/UtilTest.java
View File

@ -12,7 +12,6 @@ import org.junit.jupiter.api.Test;
import eu.dnetlib.pace.model.Person;
public class UtilTest {
static Map<String, String> params;

46
dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/collection/mag/MagUtility.scala
View File

@ -5,7 +5,17 @@ import eu.dnetlib.dhp.schema.action.AtomicAction
import eu.dnetlib.dhp.schema.common.ModelConstants
import eu.dnetlib.dhp.schema.oaf.utils.OafMapperUtils._
import eu.dnetlib.dhp.schema.oaf.utils.{OafMapperUtils, PidType}
import eu.dnetlib.dhp.schema.oaf.{Author, DataInfo, Instance, Journal, Organization, Publication, Relation, Result, Dataset => OafDataset}
import eu.dnetlib.dhp.schema.oaf.{
Author,
DataInfo,
Instance,
Journal,
Organization,
Publication,
Relation,
Result,
Dataset => OafDataset
}
import eu.dnetlib.dhp.utils.DHPUtils
import org.apache.spark.sql.types._
import org.apache.spark.sql.{Dataset, Row, SparkSession}
@ -688,33 +698,45 @@ object MagUtility extends Serializable {
o.setLegalname(field(r.getAs[String]("DisplayName"), null))
val gid = r.getAs[String]("GridId")
if (gid != null) {
o.setPid(List(
structuredProperty(gid, qualifier(
o.setPid(
List(
structuredProperty(
gid,
qualifier(
PidType.GRID.toString,
PidType.GRID.toString,
ModelConstants.DNET_PID_TYPES,
ModelConstants.DNET_PID_TYPES
),
null),
structuredProperty(r.getAs[Long]("AffiliationId").toString, qualifier(
null
),
structuredProperty(
r.getAs[Long]("AffiliationId").toString,
qualifier(
PidType.mag_id.toString,
PidType.mag_id.toString,
ModelConstants.DNET_PID_TYPES,
ModelConstants.DNET_PID_TYPES
),
null)
).asJava)
null
)
).asJava
)
} else {
o.setPid(List(
structuredProperty(r.getAs[Long]("AffiliationId").toString, qualifier(
o.setPid(
List(
structuredProperty(
r.getAs[Long]("AffiliationId").toString,
qualifier(
PidType.mag_id.toString,
PidType.mag_id.toString,
ModelConstants.DNET_PID_TYPES,
ModelConstants.DNET_PID_TYPES
),
null)
).asJava)
null
)
).asJava
)
}
val c = r.getAs[String]("Iso3166Code")
if (c != null)

6
dhp-workflows/dhp-aggregation/src/main/scala/eu/dnetlib/dhp/collection/mag/SparkMagOrganizationAS.scala
View File

@ -24,7 +24,10 @@ class SparkMagOrganizationAS (propertyPath: String, args: Array[String], log: Lo
def generateAS(spark: SparkSession, magBasePath: String, outputPath: String): Unit = {
import spark.implicits._
val organizations = MagUtility.loadMagEntity(spark, "Affiliations", magBasePath)
organizations.map(r => MagUtility.generateOrganization(r)).write.mode(SaveMode.Overwrite)
organizations
.map(r => MagUtility.generateOrganization(r))
.write
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.text(outputPath)
}
@ -33,6 +36,7 @@ class SparkMagOrganizationAS (propertyPath: String, args: Array[String], log: Lo
object SparkMagOrganizationAS {
val log: Logger = LoggerFactory.getLogger(SparkMagOrganizationAS.getClass)
def main(args: Array[String]): Unit = {
new SparkMagOrganizationAS("/eu/dnetlib/dhp/collection/mag/create_organization_AS.json", args, log)
.initialize()

1
dhp-workflows/dhp-aggregation/src/test/java/eu/dnetlib/dhp/collection/plugin/rest/OsfPreprintCollectorTest.java
View File

@ -1,3 +1,4 @@
package eu.dnetlib.dhp.collection.plugin.rest;
import java.util.HashMap;

1
dhp-workflows/dhp-aggregation/src/test/scala/eu/dnetlib/dhp/collection/mag/MAGMappingTest.scala
View File

@ -10,7 +10,6 @@ class MAGMappingTest {
val mapper = new ObjectMapper()
def mappingTest(): Unit = {
val spark = SparkSession

1
dhp-workflows/dhp-dedup-openaire/src/test/java/eu/dnetlib/dhp/oa/dedup/SparkDedupTest.java
View File

@ -258,7 +258,6 @@ public class SparkDedupTest implements Serializable {
assertEquals(115, sw_simrel.count());
}
// check if the first relation in the whitelist exists
assertTrue(
sw_simrel

1
dhp-workflows/dhp-doiboost/src/test/java/eu/dnetlib/doiboost/orcid/OrcidClientTest.java
View File

@ -31,7 +31,6 @@ import eu.dnetlib.dhp.schema.orcid.AuthorData;
import eu.dnetlib.doiboost.orcid.util.DownloadsReport;
import eu.dnetlib.doiboost.orcid.util.MultiAttemptsHttpConnector;
public class OrcidClientTest {
final int REQ_LIMIT = 24;
final int REQ_MAX_TEST = 100;

3
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/raw/GenerateEntitiesApplication.java
View File

@ -130,7 +130,8 @@ public class GenerateEntitiesApplication extends AbstractMigrationApplication {
switch (mode) {
case claim:
save(
inputRdd.keyBy(oaf -> ModelSupport.idFn().apply(oaf))
inputRdd
.keyBy(oaf -> ModelSupport.idFn().apply(oaf))
.groupByKey()
.map(t -> MergeUtils.mergeGroup(t._1, t._2.iterator())),
// .mapToPair(oaf -> new Tuple2<>(ModelSupport.idFn().apply(oaf), oaf))