|
|
@ -2,12 +2,11 @@ package eu.dnetlib.dhp.sx.graph
|
|
|
|
|
|
|
|
|
|
|
|
import com.fasterxml.jackson.databind.ObjectMapper
|
|
|
|
import com.fasterxml.jackson.databind.ObjectMapper
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser
|
|
|
|
import eu.dnetlib.dhp.application.ArgumentApplicationParser
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.{OtherResearchProduct, Publication, Relation, Software,Dataset => OafDataset}
|
|
|
|
import eu.dnetlib.dhp.schema.oaf.{OtherResearchProduct, Publication, Relation, Result, Software, Dataset => OafDataset}
|
|
|
|
import org.apache.commons.io.IOUtils
|
|
|
|
import org.apache.commons.io.IOUtils
|
|
|
|
import org.apache.spark.SparkConf
|
|
|
|
import org.apache.spark.SparkConf
|
|
|
|
import org.apache.spark.sql.{Encoder, Encoders, SaveMode, SparkSession}
|
|
|
|
import org.apache.spark.sql.{Encoder, Encoders, SaveMode, SparkSession}
|
|
|
|
import org.slf4j.{Logger, LoggerFactory}
|
|
|
|
import org.slf4j.{Logger, LoggerFactory}
|
|
|
|
|
|
|
|
|
|
|
|
object SparkConvertRDDtoDataset {
|
|
|
|
object SparkConvertRDDtoDataset {
|
|
|
|
|
|
|
|
|
|
|
|
def main(args: Array[String]): Unit = {
|
|
|
|
def main(args: Array[String]): Unit = {
|
|
|
@ -32,39 +31,40 @@ object SparkConvertRDDtoDataset {
|
|
|
|
val entityPath = s"$t/entities"
|
|
|
|
val entityPath = s"$t/entities"
|
|
|
|
val relPath = s"$t/relation"
|
|
|
|
val relPath = s"$t/relation"
|
|
|
|
val mapper = new ObjectMapper()
|
|
|
|
val mapper = new ObjectMapper()
|
|
|
|
implicit val datasetEncoder: Encoder[OafDataset] = Encoders.kryo(classOf[OafDataset])
|
|
|
|
implicit val datasetEncoder: Encoder[OafDataset] = Encoders.kryo(classOf[OafDataset])
|
|
|
|
implicit val publicationEncoder: Encoder[Publication] = Encoders.kryo(classOf[Publication])
|
|
|
|
implicit val publicationEncoder: Encoder[Publication] = Encoders.kryo(classOf[Publication])
|
|
|
|
implicit val relationEncoder: Encoder[Relation] = Encoders.kryo(classOf[Relation])
|
|
|
|
implicit val relationEncoder: Encoder[Relation] = Encoders.kryo(classOf[Relation])
|
|
|
|
implicit val orpEncoder: Encoder[OtherResearchProduct] = Encoders.kryo(classOf[OtherResearchProduct])
|
|
|
|
implicit val orpEncoder: Encoder[OtherResearchProduct] = Encoders.kryo(classOf[OtherResearchProduct])
|
|
|
|
implicit val softwareEncoder: Encoder[Software] = Encoders.kryo(classOf[Software])
|
|
|
|
implicit val softwareEncoder: Encoder[Software] = Encoders.kryo(classOf[Software])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
log.info("Converting dataset")
|
|
|
|
log.info("Converting dataset")
|
|
|
|
val rddDataset = spark.sparkContext.textFile(s"$sourcePath/dataset").map(s => mapper.readValue(s, classOf[OafDataset]))
|
|
|
|
val rddDataset =spark.sparkContext.textFile(s"$sourcePath/dataset").map(s => mapper.readValue(s, classOf[OafDataset])).filter(r=> r.getDataInfo!= null && r.getDataInfo.getDeletedbyinference == false)
|
|
|
|
spark.createDataset(rddDataset).as[OafDataset].write.mode(SaveMode.Overwrite).save(s"$entityPath/dataset")
|
|
|
|
spark.createDataset(rddDataset).as[OafDataset].write.mode(SaveMode.Overwrite).save(s"$entityPath/dataset")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
log.info("Converting publication")
|
|
|
|
log.info("Converting publication")
|
|
|
|
val rddPublication = spark.sparkContext.textFile(s"$sourcePath/publication").map(s => mapper.readValue(s, classOf[Publication]))
|
|
|
|
val rddPublication =spark.sparkContext.textFile(s"$sourcePath/publication").map(s => mapper.readValue(s, classOf[Publication])).filter(r=> r.getDataInfo!= null && r.getDataInfo.getDeletedbyinference == false)
|
|
|
|
spark.createDataset(rddPublication).as[Publication].write.mode(SaveMode.Overwrite).save(s"$entityPath/publication")
|
|
|
|
spark.createDataset(rddPublication).as[Publication].write.mode(SaveMode.Overwrite).save(s"$entityPath/publication")
|
|
|
|
|
|
|
|
|
|
|
|
log.info("Converting software")
|
|
|
|
log.info("Converting software")
|
|
|
|
val rddSoftware = spark.sparkContext.textFile(s"$sourcePath/software").map(s => mapper.readValue(s, classOf[Software]))
|
|
|
|
val rddSoftware =spark.sparkContext.textFile(s"$sourcePath/software").map(s => mapper.readValue(s, classOf[Software])).filter(r=> r.getDataInfo!= null && r.getDataInfo.getDeletedbyinference == false)
|
|
|
|
spark.createDataset(rddSoftware).as[Software].write.mode(SaveMode.Overwrite).save(s"$entityPath/software")
|
|
|
|
spark.createDataset(rddSoftware).as[Software].write.mode(SaveMode.Overwrite).save(s"$entityPath/software")
|
|
|
|
|
|
|
|
|
|
|
|
log.info("Converting otherresearchproduct")
|
|
|
|
log.info("Converting otherresearchproduct")
|
|
|
|
val rddOtherResearchProduct = spark.sparkContext.textFile(s"$sourcePath/otherresearchproduct").map(s => mapper.readValue(s, classOf[OtherResearchProduct]))
|
|
|
|
val rddOtherResearchProduct =spark.sparkContext.textFile(s"$sourcePath/otherresearchproduct").map(s => mapper.readValue(s, classOf[OtherResearchProduct])).filter(r=> r.getDataInfo!= null && r.getDataInfo.getDeletedbyinference == false)
|
|
|
|
spark.createDataset(rddOtherResearchProduct).as[OtherResearchProduct].write.mode(SaveMode.Overwrite).save(s"$entityPath/otherresearchproduct")
|
|
|
|
spark.createDataset(rddOtherResearchProduct).as[OtherResearchProduct].write.mode(SaveMode.Overwrite).save(s"$entityPath/otherresearchproduct")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
log.info("Converting Relation")
|
|
|
|
log.info("Converting Relation")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
val relationSemanticFilter = List("cites", "iscitedby", "merges", "ismergedin")
|
|
|
|
val relationSemanticFilter = List("cites", "iscitedby","merges", "ismergedin", "HasAmongTopNSimilarDocuments","IsAmongTopNSimilarDocuments" )
|
|
|
|
|
|
|
|
|
|
|
|
val rddRelation = spark.sparkContext.textFile(s"$sourcePath/relation")
|
|
|
|
val rddRelation =spark.sparkContext.textFile(s"$sourcePath/relation")
|
|
|
|
.map(s => mapper.readValue(s, classOf[Relation]))
|
|
|
|
.map(s => mapper.readValue(s, classOf[Relation]))
|
|
|
|
.filter(r => r.getSource.startsWith("50") && r.getTarget.startsWith("50"))
|
|
|
|
.filter(r=> r.getDataInfo!= null && r.getDataInfo.getDeletedbyinference == false)
|
|
|
|
|
|
|
|
.filter(r=> r.getSource.startsWith("50") && r.getTarget.startsWith("50"))
|
|
|
|
.filter(r => !relationSemanticFilter.exists(k => k.equalsIgnoreCase(r.getRelClass)))
|
|
|
|
.filter(r => !relationSemanticFilter.exists(k => k.equalsIgnoreCase(r.getRelClass)))
|
|
|
|
spark.createDataset(rddRelation).as[Relation].write.mode(SaveMode.Overwrite).save(s"$relPath")
|
|
|
|
spark.createDataset(rddRelation).as[Relation].write.mode(SaveMode.Overwrite).save(s"$relPath")
|
|
|
|
|
|
|
|
|
|
|
|