From 6efab4d88e7ce481896e5569e1801daf81c96777 Mon Sep 17 00:00:00 2001 From: Sandro La Bruzzo Date: Thu, 16 May 2024 16:19:18 +0200 Subject: [PATCH] fixed scholexplorer bug --- .../dhp/sx/graph/scholix/ScholixUtils.scala | 2 +- dhp-shade-package/pom.xml | 150 +++++++++--------- .../dhp/sx/graph/ScholexplorerUtils.scala | 15 +- .../graph/SparkCreateScholexplorerDump.scala | 23 ++- .../graph/scholix/ScholixGenerationTest.scala | 17 +- 5 files changed, 112 insertions(+), 95 deletions(-) diff --git a/dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixUtils.scala b/dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixUtils.scala index f256ca1a12..72a17777e9 100644 --- a/dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixUtils.scala +++ b/dhp-common/src/main/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixUtils.scala @@ -389,7 +389,7 @@ object ScholixUtils extends Serializable { if (persistentIdentifiers.isEmpty) return null s.setLocalIdentifier(persistentIdentifiers.asJava) - s.setTypology(r.getResulttype.getClassid) +// s.setTypology(r.getResulttype.getClassid) s.setSubType(r.getInstance().get(0).getInstancetype.getClassname) diff --git a/dhp-shade-package/pom.xml b/dhp-shade-package/pom.xml index 128a571165..fd9c040660 100644 --- a/dhp-shade-package/pom.xml +++ b/dhp-shade-package/pom.xml @@ -31,86 +31,86 @@ dhp-actionmanager ${project.version} - - eu.dnetlib.dhp - dhp-aggregation - ${project.version} - - - eu.dnetlib.dhp - dhp-blacklist - ${project.version} - - - eu.dnetlib.dhp - dhp-broker-events - ${project.version} - - - eu.dnetlib.dhp - dhp-dedup-openaire - ${project.version} - - - eu.dnetlib.dhp - dhp-enrichment - ${project.version} - + + + + + + + + + + + + + + + + + + + + + + + + + eu.dnetlib.dhp dhp-graph-mapper ${project.version} - - eu.dnetlib.dhp - dhp-graph-provision - ${project.version} - - - eu.dnetlib.dhp - dhp-impact-indicators - ${project.version} - - - eu.dnetlib.dhp - dhp-stats-actionsets - ${project.version} - - - eu.dnetlib.dhp - dhp-stats-hist-snaps - ${project.version} - - - eu.dnetlib.dhp - dhp-stats-monitor-irish - ${project.version} - - - eu.dnetlib.dhp - dhp-stats-promote - ${project.version} - - - eu.dnetlib.dhp - dhp-stats-update - ${project.version} - - - eu.dnetlib.dhp - dhp-swh - ${project.version} - - - eu.dnetlib.dhp - dhp-usage-raw-data-update - ${project.version} - - - eu.dnetlib.dhp - dhp-usage-stats-build - ${project.version} - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/ScholexplorerUtils.scala b/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/ScholexplorerUtils.scala index 95564d5236..f62f271e30 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/ScholexplorerUtils.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/ScholexplorerUtils.scala @@ -1,14 +1,8 @@ package eu.dnetlib.dhp.sx.graph +import com.fasterxml.jackson.databind.ObjectMapper import eu.dnetlib.dhp.schema.oaf.{KeyValue, Result, StructuredProperty} -import eu.dnetlib.dhp.schema.sx.scholix.{ - Scholix, - ScholixCollectedFrom, - ScholixEntityId, - ScholixIdentifier, - ScholixRelationship, - ScholixResource -} +import eu.dnetlib.dhp.schema.sx.scholix.{Scholix, ScholixCollectedFrom, ScholixEntityId, ScholixIdentifier, ScholixRelationship, ScholixResource} import org.json4s import org.json4s.DefaultFormats import org.json4s.jackson.JsonMethods.parse @@ -28,6 +22,7 @@ case class RelKeyValue(key: String, value: String) {} object ScholexplorerUtils { val OPENAIRE_IDENTIFIER_SCHEMA: String = "OpenAIRE Identifier" + val mapper= new ObjectMapper() case class RelationVocabulary(original: String, inverse: String) {} @@ -242,7 +237,7 @@ object ScholexplorerUtils { s } - def updateTarget(s: Scholix, t: ScholixResource): Scholix = { + def updateTarget(s: Scholix, t: ScholixResource): String = { s.setTarget(t) val spublishers: Seq[ScholixEntityId] = @@ -251,6 +246,6 @@ object ScholexplorerUtils { if (t.getPublisher != null && !t.getPublisher.isEmpty) t.getPublisher.asScala else List() val mergedPublishers = spublishers.union(tpublishers).distinct.take(10).toList s.setPublisher(mergedPublishers.asJava) - s + mapper.writeValueAsString(s) } } diff --git a/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/SparkCreateScholexplorerDump.scala b/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/SparkCreateScholexplorerDump.scala index 1211dcc786..32aa686659 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/SparkCreateScholexplorerDump.scala +++ b/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/SparkCreateScholexplorerDump.scala @@ -11,7 +11,7 @@ import eu.dnetlib.dhp.schema.oaf.{ Dataset => OafDataset } import eu.dnetlib.dhp.schema.sx.scholix.{Scholix, ScholixResource} -import org.apache.spark.sql.functions.{col, concat, expr, md5} +import org.apache.spark.sql.functions.{col, concat, expr, first, md5} import org.apache.spark.sql.types.StructType import org.apache.spark.sql._ import org.slf4j.{Logger, LoggerFactory} @@ -89,7 +89,13 @@ class SparkCreateScholexplorerDump(propertyPath: String, args: Array[String], lo .withColumn("cf", expr("transform(collectedfrom, x -> struct(x.key, x.value))")) .drop("collectedfrom") .withColumnRenamed("cf", "collectedfrom") - .distinct() + .groupBy(col("id")) + .agg( + first("source").alias("source"), + first("target").alias("target"), + first("relClass").alias("relClass"), + first("collectedfrom").alias("collectedfrom") + ) bidRel.write.mode(SaveMode.Overwrite).save(s"$otuputPath/relation") @@ -97,27 +103,32 @@ class SparkCreateScholexplorerDump(propertyPath: String, args: Array[String], lo def generateScholix(outputPath: String, spark: SparkSession): Unit = { implicit val scholixResourceEncoder: Encoder[ScholixResource] = Encoders.bean(classOf[ScholixResource]) - implicit val scholixEncoder: Encoder[Scholix] = Encoders.bean(classOf[Scholix]) + implicit val scholixEncoder: Encoder[Scholix] = Encoders.kryo(classOf[Scholix]) import spark.implicits._ val relations = spark.read.load(s"$outputPath/relation").as[RelationInfo] val resource = spark.read.load(s"$outputPath/resource").as[ScholixResource] + + val scholix_one_verse = relations .joinWith(resource, relations("source") === resource("dnetIdentifier"), "inner") .map(res => ScholexplorerUtils.generateScholix(res._1, res._2)) + .map(s=> (s.getIdentifier, s))(Encoders.tuple(Encoders.STRING, Encoders.kryo(classOf[Scholix]))) + val resourceTarget = relations .joinWith(resource, relations("target") === resource("dnetIdentifier"), "inner") .map(res => (res._1.id, res._2))(Encoders.tuple(Encoders.STRING, Encoders.kryo(classOf[ScholixResource]))) + scholix_one_verse - .joinWith(resourceTarget, scholix_one_verse("identifier") === resourceTarget("_1"), "inner") - .map(k => ScholexplorerUtils.updateTarget(k._1, k._2._2)) + .joinWith(resourceTarget, scholix_one_verse("_1") === resourceTarget("_1"), "inner") + .map(k => ScholexplorerUtils.updateTarget(k._1._2, k._2._2)) .write .mode(SaveMode.Overwrite) .option("compression", "gzip") - .json(s"$outputPath/scholix") + .text(s"$outputPath/scholix") } } diff --git a/dhp-workflows/dhp-graph-mapper/src/test/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixGenerationTest.scala b/dhp-workflows/dhp-graph-mapper/src/test/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixGenerationTest.scala index 0a2872cb48..67d40dcf12 100644 --- a/dhp-workflows/dhp-graph-mapper/src/test/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixGenerationTest.scala +++ b/dhp-workflows/dhp-graph-mapper/src/test/scala/eu/dnetlib/dhp/sx/graph/scholix/ScholixGenerationTest.scala @@ -1,17 +1,28 @@ package eu.dnetlib.dhp.sx.graph.scholix +import eu.dnetlib.dhp.schema.sx.scholix.ScholixResource import eu.dnetlib.dhp.sx.graph.SparkCreateScholexplorerDump -import org.apache.spark.sql.SparkSession +import org.apache.spark.SparkConf +import org.apache.spark.sql.{Encoder, Encoders, SparkSession} import org.junit.jupiter.api.Test +import org.objenesis.strategy.StdInstantiatorStrategy class ScholixGenerationTest { @Test def generateScholix(): Unit = { + val spark: SparkSession = SparkSession.builder().master("local[*]").getOrCreate() val app = new SparkCreateScholexplorerDump(null, null, null) -// app.generateScholixResource("/home/sandro/Downloads/scholix_sample/", "/home/sandro/Downloads/scholix/", spark) -// app.generateBidirectionalRelations("/home/sandro/Downloads/scholix_sample/", "/home/sandro/Downloads/scholix/", spark) +// app.generateScholixResource("/home/sandro/Downloads/scholix_sample/", "/home/sandro/Downloads/scholix/", spark) +// app.generateBidirectionalRelations( +// "/home/sandro/Downloads/scholix_sample/", +// "/home/sandro/Downloads/scholix/", +// spark +// ) app.generateScholix("/home/sandro/Downloads/scholix/", spark) + + + } }