dnet-hadoop/dhp-workflows/dhp-graph-mapper/src/main/scala/eu/dnetlib/dhp/sx/graph/SparkCreateInputGraph.scala

145 lines
4.1 KiB
Scala

package eu.dnetlib.dhp.sx.graph
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.schema.oaf.{Dataset => OafDataset, _}
import org.apache.commons.io.IOUtils
import org.apache.spark.SparkConf
import org.apache.spark.sql._
import org.slf4j.{Logger, LoggerFactory}
object SparkCreateInputGraph {
def main(args: Array[String]): Unit = {
val log: Logger = LoggerFactory.getLogger(getClass)
val conf: SparkConf = new SparkConf()
val parser = new ArgumentApplicationParser(
IOUtils.toString(
getClass.getResourceAsStream("/eu/dnetlib/dhp/sx/graph/extract_entities_params.json")
)
)
parser.parseArgument(args)
val spark: SparkSession =
SparkSession
.builder()
.config(conf)
.appName(getClass.getSimpleName)
.master(parser.get("master"))
.getOrCreate()
val resultObject = List(
("publication", classOf[Publication]),
("dataset", classOf[OafDataset]),
("software", classOf[Software]),
("otherResearchProduct", classOf[OtherResearchProduct])
)
implicit val oafEncoder: Encoder[Oaf] = Encoders.kryo(classOf[Oaf])
implicit val publicationEncoder: Encoder[Publication] = Encoders.kryo(classOf[Publication])
implicit val datasetEncoder: Encoder[OafDataset] = Encoders.kryo(classOf[OafDataset])
implicit val softwareEncoder: Encoder[Software] = Encoders.kryo(classOf[Software])
implicit val orpEncoder: Encoder[OtherResearchProduct] =
Encoders.kryo(classOf[OtherResearchProduct])
implicit val relEncoder: Encoder[Relation] = Encoders.kryo(classOf[Relation])
val sourcePath = parser.get("sourcePath")
log.info(s"sourcePath -> $sourcePath")
val targetPath = parser.get("targetPath")
log.info(s"targetPath -> $targetPath")
val oafDs: Dataset[Oaf] = spark.read.load(s"$sourcePath/*").as[Oaf]
log.info("Extract Publication")
oafDs
.filter(o => o.isInstanceOf[Publication])
.map(p => p.asInstanceOf[Publication])
.write
.mode(SaveMode.Overwrite)
.save(s"$targetPath/extracted/publication")
log.info("Extract dataset")
oafDs
.filter(o => o.isInstanceOf[OafDataset])
.map(p => p.asInstanceOf[OafDataset])
.write
.mode(SaveMode.Overwrite)
.save(s"$targetPath/extracted/dataset")
log.info("Extract software")
oafDs
.filter(o => o.isInstanceOf[Software])
.map(p => p.asInstanceOf[Software])
.write
.mode(SaveMode.Overwrite)
.save(s"$targetPath/extracted/software")
log.info("Extract otherResearchProduct")
oafDs
.filter(o => o.isInstanceOf[OtherResearchProduct])
.map(p => p.asInstanceOf[OtherResearchProduct])
.write
.mode(SaveMode.Overwrite)
.save(s"$targetPath/extracted/otherResearchProduct")
log.info("Extract Relation")
oafDs
.filter(o => o.isInstanceOf[Relation])
.map(p => p.asInstanceOf[Relation])
.write
.mode(SaveMode.Overwrite)
.save(s"$targetPath/extracted/relation")
resultObject.foreach { r =>
log.info(s"Make ${r._1} unique")
makeDatasetUnique(
s"$targetPath/extracted/${r._1}",
s"$targetPath/preprocess/${r._1}",
spark,
r._2
)
}
}
def extractEntities[T <: Oaf](
oafDs: Dataset[Oaf],
targetPath: String,
clazz: Class[T],
log: Logger
): Unit = {
implicit val resEncoder: Encoder[T] = Encoders.kryo(clazz)
log.info(s"Extract ${clazz.getSimpleName}")
oafDs
.filter(o => o.isInstanceOf[T])
.map(p => p.asInstanceOf[T])
.write
.mode(SaveMode.Overwrite)
.save(targetPath)
}
def makeDatasetUnique[T <: Result](
sourcePath: String,
targetPath: String,
spark: SparkSession,
clazz: Class[T]
): Unit = {
import spark.implicits._
implicit val resEncoder: Encoder[T] = Encoders.kryo(clazz)
val ds: Dataset[T] = spark.read.load(sourcePath).as[T]
ds.groupByKey(_.getId)
.reduceGroups { (x, y) =>
x.mergeFrom(y)
x
}
.map(_._2)
.write
.mode(SaveMode.Overwrite)
.save(targetPath)
}
}