Hosted By Map - first attempt for the creation of intermedia information to be used to applu the hosted by map on the graph entities

This commit is contained in:
Miriam Baglioni 2021-07-30 17:56:27 +02:00
parent d8b9b0553b
commit 7c6ea2f4c7
2 changed files with 225 additions and 35 deletions

View File

@ -1,5 +1,6 @@
package eu.dnetlib.dhp.oa.graph.hostedbymap package eu.dnetlib.dhp.oa.graph.hostedbymap
import eu.dnetlib.dhp.oa.graph.hostedbymap.model.EntityInfo
import org.apache.spark.sql.{Dataset, Encoder, Encoders, TypedColumn} import org.apache.spark.sql.{Dataset, Encoder, Encoders, TypedColumn}
import org.apache.spark.sql.expressions.Aggregator import org.apache.spark.sql.expressions.Aggregator
@ -25,43 +26,10 @@ object Aggregators {
} }
def createHostedByItemTypes(df: Dataset[HostedByItemType]): Dataset[HostedByItemType] = {
val transformedData : Dataset[HostedByItemType] = df
.groupByKey(_.id)(Encoders.STRING)
.agg(Aggregators.hostedByAggregator)
.map{
case (id:String , res:HostedByItemType) => res
}(Encoders.product[HostedByItemType])
transformedData
}
val hostedByAggregator: TypedColumn[HostedByItemType, HostedByItemType] = new Aggregator[HostedByItemType, HostedByItemType, HostedByItemType] {
override def zero: HostedByItemType = HostedByItemType("","","","","",false)
override def reduce(b: HostedByItemType, a:HostedByItemType): HostedByItemType = {
return merge(b, a)
}
override def merge(b1: HostedByItemType, b2: HostedByItemType): HostedByItemType = {
if (b1 == null){
return b2
}
if(b2 == null){
return b1
}
HostedByItemType(getId(b1.id, b2.id), getId(b1.officialname, b2.officialname), getId(b1.issn, b2.issn), getId(b1.eissn, b2.eissn), getId(b1.lissn, b2.lissn), b1.openAccess || b2.openAccess)
}
override def finish(reduction: HostedByItemType): HostedByItemType = reduction
override def bufferEncoder: Encoder[HostedByItemType] = Encoders.product[HostedByItemType]
override def outputEncoder: Encoder[HostedByItemType] = Encoders.product[HostedByItemType]
}.toColumn
def explodeHostedByItemType(df: Dataset[(String, HostedByItemType)]): Dataset[(String, HostedByItemType)] = { def explodeHostedByItemType(df: Dataset[(String, HostedByItemType)]): Dataset[(String, HostedByItemType)] = {
val transformedData : Dataset[(String, HostedByItemType)] = df val transformedData : Dataset[(String, HostedByItemType)] = df
.groupByKey(_._1)(Encoders.STRING) .groupByKey(_._1)(Encoders.STRING)
.agg(Aggregators.hostedByAggregator1) .agg(Aggregators.hostedByAggregator)
.map{ .map{
case (id:String , res:(String, HostedByItemType)) => res case (id:String , res:(String, HostedByItemType)) => res
}(Encoders.tuple(Encoders.STRING, Encoders.product[HostedByItemType])) }(Encoders.tuple(Encoders.STRING, Encoders.product[HostedByItemType]))
@ -69,7 +37,7 @@ object Aggregators {
transformedData transformedData
} }
val hostedByAggregator1: TypedColumn[(String, HostedByItemType), (String, HostedByItemType)] = new Aggregator[(String, HostedByItemType), (String, HostedByItemType), (String, HostedByItemType)] { val hostedByAggregator: TypedColumn[(String, HostedByItemType), (String, HostedByItemType)] = new Aggregator[(String, HostedByItemType), (String, HostedByItemType), (String, HostedByItemType)] {
override def zero: (String, HostedByItemType) = ("", HostedByItemType("","","","","",false)) override def zero: (String, HostedByItemType) = ("", HostedByItemType("","","","","",false))
override def reduce(b: (String, HostedByItemType), a:(String,HostedByItemType)): (String, HostedByItemType) = { override def reduce(b: (String, HostedByItemType), a:(String,HostedByItemType)): (String, HostedByItemType) = {
return merge(b, a) return merge(b, a)
@ -94,4 +62,79 @@ object Aggregators {
override def outputEncoder: Encoder[(String,HostedByItemType)] = Encoders.tuple(Encoders.STRING,Encoders.product[HostedByItemType]) override def outputEncoder: Encoder[(String,HostedByItemType)] = Encoders.tuple(Encoders.STRING,Encoders.product[HostedByItemType])
}.toColumn }.toColumn
def hostedByToSingleDSId(df: Dataset[ HostedByItemType]): Dataset[ HostedByItemType] = {
val transformedData : Dataset[HostedByItemType] = df
.groupByKey(_.id)(Encoders.STRING)
.agg(Aggregators.hostedByToDSAggregator)
.map{
case (id:String , res: HostedByItemType) => res
}(Encoders.product[HostedByItemType])
transformedData
}
def hostedByToDSAggregator: TypedColumn[HostedByItemType, HostedByItemType] = new Aggregator[HostedByItemType, HostedByItemType, HostedByItemType] {
override def zero: HostedByItemType = HostedByItemType("","","","","",false)
override def reduce(b: HostedByItemType, a:HostedByItemType): HostedByItemType = {
return merge(b, a)
}
override def merge(b1: HostedByItemType, b2: HostedByItemType): HostedByItemType = {
if (b1 == null){
return b2
}
if(b2 == null){
return b1
}
if(!b1.id.equals("")){
return HostedByItemType(b1.id, b1.officialname, b1.issn, b1.eissn, b1.lissn, b1.openAccess || b2.openAccess)
}
return HostedByItemType(b2.id, b2.officialname, b2.issn, b2.eissn, b2.lissn, b1.openAccess || b2.openAccess)
}
override def finish(reduction: HostedByItemType): HostedByItemType = reduction
override def bufferEncoder: Encoder[HostedByItemType] = Encoders.product[HostedByItemType]
override def outputEncoder: Encoder[HostedByItemType] = Encoders.product[HostedByItemType]
}.toColumn
def resultToSingleIdAggregator: TypedColumn[EntityInfo, EntityInfo] = new Aggregator[EntityInfo, EntityInfo, EntityInfo]{
override def zero: EntityInfo = EntityInfo.newInstance("","","")
override def reduce(b: EntityInfo, a:EntityInfo): EntityInfo = {
return merge(b, a)
}
override def merge(b1: EntityInfo, b2: EntityInfo): EntityInfo = {
if (b1 == null){
return b2
}
if(b2 == null){
return b1
}
if(!b1.getHb_id.equals("")){
b1.setOpenaccess(b1.getOpenaccess || b2.getOpenaccess)
}
b2.setOpenaccess(b1.getOpenaccess || b2.getOpenaccess)
b2
}
override def finish(reduction: EntityInfo): EntityInfo = reduction
override def bufferEncoder: Encoder[EntityInfo] = Encoders.bean(classOf[EntityInfo])
override def outputEncoder: Encoder[EntityInfo] = Encoders.bean(classOf[EntityInfo])
}.toColumn
def resultToSingleId(df:Dataset[EntityInfo]): Dataset[EntityInfo] = {
val transformedData : Dataset[EntityInfo] = df
.groupByKey(_.getId)(Encoders.STRING)
.agg(Aggregators.resultToSingleIdAggregator)
.map{
case (id:String , res: EntityInfo) => res
}(Encoders.bean(classOf[EntityInfo]))
transformedData
}
} }

View File

@ -0,0 +1,147 @@
package eu.dnetlib.dhp.oa.graph.hostedbymap
import com.fasterxml.jackson.databind.ObjectMapper
import eu.dnetlib.dhp.application.ArgumentApplicationParser
import eu.dnetlib.dhp.oa.graph.hostedbymap.model.{DatasourceInfo, EntityInfo}
import eu.dnetlib.dhp.schema.oaf.{Datasource, Journal, Publication}
import org.apache.commons.io.IOUtils
import org.apache.spark.SparkConf
import org.apache.spark.sql.{Dataset, Encoder, Encoders, SaveMode, SparkSession}
import org.json4s
import org.json4s.DefaultFormats
import org.json4s.jackson.JsonMethods.parse
import org.slf4j.{Logger, LoggerFactory}
object SparkPrepareHostedByInfoToApply {
implicit val mapEncoderDSInfo: Encoder[DatasourceInfo] = Encoders.kryo[DatasourceInfo]
implicit val mapEncoderPInfo: Encoder[EntityInfo] = Encoders.kryo[EntityInfo]
def getList(id: String, j: Journal, name: String ) : List[EntityInfo] = {
var lst:List[EntityInfo] = List()
if (j.getIssnLinking != null && !j.getIssnLinking.equals("")){
lst = EntityInfo.newInstance(id, j.getIssnLinking, name) :: lst
}
if (j.getIssnOnline != null && !j.getIssnOnline.equals("")){
lst = EntityInfo.newInstance(id, j.getIssnOnline, name) :: lst
}
if (j.getIssnPrinted != null && !j.getIssnPrinted.equals("")){
lst = EntityInfo.newInstance(id, j.getIssnPrinted, name) :: lst
}
lst
}
def prepareResultInfo(spark:SparkSession, publicationPath:String) : Dataset[EntityInfo] = {
implicit val mapEncoderPubs: Encoder[Publication] = Encoders.bean(classOf[Publication])
val mapper = new ObjectMapper()
val dd : Dataset[Publication] = spark.read.textFile(publicationPath)
.map(r => mapper.readValue(r, classOf[Publication]))
dd.filter(p => p.getJournal != null ).flatMap(p => getList(p.getId, p.getJournal, ""))
}
def prepareDatasourceInfo(spark:SparkSession, datasourcePath:String) : Dataset[DatasourceInfo] = {
implicit val mapEncoderDats: Encoder[Datasource] = Encoders.bean(classOf[Datasource])
val mapper = new ObjectMapper()
val dd : Dataset[Datasource] = spark.read.textFile(datasourcePath)
.map(r => mapper.readValue(r, classOf[Datasource]))
dd.filter(d => d.getJournal != null ).map(d => DatasourceInfo.newInstance(d.getId, d.getOfficialname.getValue,
d.getJournal.getIssnPrinted, d.getJournal.getIssnOnline, d.getJournal.getIssnLinking))
}
def toHostedByItem(input:String): HostedByItemType = {
implicit lazy val formats: DefaultFormats.type = org.json4s.DefaultFormats
lazy val json: json4s.JValue = parse(input)
val c :Map[String,HostedByItemType] = json.extract[Map[String, HostedByItemType]]
c.values.head
}
def explodeJournalInfo(input: DatasourceInfo): List[EntityInfo] = {
var lst : List[EntityInfo] = List()
if (input.getEissn != null && !input.getEissn.equals("")){
lst = EntityInfo.newInstance(input.getId, input.getEissn, input.getOfficialname, input.getOpenAccess) :: lst
}
lst
}
def main(args: Array[String]): Unit = {
val logger: Logger = LoggerFactory.getLogger(getClass)
val conf: SparkConf = new SparkConf()
val parser = new ArgumentApplicationParser(IOUtils.toString(getClass.getResourceAsStream("/eu/dnetlib/dhp/oa/graph/hostedbymap/hostedby_prepare_params.json")))
parser.parseArgument(args)
val spark: SparkSession =
SparkSession
.builder()
.config(conf)
.appName(getClass.getSimpleName)
.master(parser.get("master")).getOrCreate()
val graphPath = parser.get("graphPath")
val outputPath = parser.get("outputPath")
val hostedByMapPath = parser.get("hostedByMapPath")
implicit val formats = DefaultFormats
logger.info("Getting the Datasources")
import spark.implicits._
//STEP1: leggere le DS e creare le entries {dsid, dsofficialname, issn, eissn, lissn, openaccess}
val datasourceInfoDataset: Dataset[DatasourceInfo] = prepareDatasourceInfo(spark, "$graphPath/datasource")
//STEP2: leggere la hostedbymap e raggruppare per datasource id
val hostedByDataset = Aggregators.hostedByToSingleDSId(spark.createDataset(spark.sparkContext.textFile(hostedByMapPath).map(toHostedByItem)))
//STEP3: eseguire una join fra le datasource e la hostedby map (left) per settare se la datasource e' open access o no
//ed esplodere l'info della datasource per ogni journal id diverso da nullo
val join : Dataset[EntityInfo] = datasourceInfoDataset.joinWith(hostedByDataset,
datasourceInfoDataset.col("id").equalTo(hostedByDataset.col("id"), "left"))
.map(t2 => {
val dsi : DatasourceInfo = t2._1
if(t2._2 != null){
dsi.setOpenAccess(t2._2.openAccess)
}
dsi
}).flatMap(explodeJournalInfo)
//STEP4: creare la mappa publication id issn, eissn, lissn esplosa
val resultInfoDataset:Dataset[EntityInfo] = prepareResultInfo(spark, "$graphPath/publication")
//STEP5: join di join con resultInfo sul journal_id dal result con left
// e riduzione di tutti i result con lo stesso id in una unica entry
Aggregators.resultToSingleId(resultInfoDataset.joinWith(join, resultInfoDataset.col("journal_id").equalTo(join.col("journal_id")), "left")
.map(t2 => {
val res: EntityInfo = t2._1
if(t2._2 != null ){
val ds = t2._2
res.setHb_id(ds.getId)
res.setOpenaccess(ds.getOpenaccess)
res.setName(ds.getName)
}
res
})).write.mode(SaveMode.Overwrite).option("compression", "gzip").json(outputPath)
}
}