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

minor fix

This commit is contained in:
Sandro La Bruzzo 2021-07-25 12:06:41 +02:00
parent 8fac10c91e
commit 848aabbb6c
2 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkCreateScholix.scala
View File

@ -42,6 +42,7 @@ object SparkCreateScholix {
val relationDS: Dataset[(String, Relation)] = spark.read.load(relationPath).as[Relation]
.filter(r => (r.getDataInfo== null || r.getDataInfo.getDeletedbyinference == false) && !r.getRelClass.toLowerCase.contains("merge"))
.map(r => (r.getSource, r))(Encoders.tuple(Encoders.STRING, relEncoder))
val summaryDS: Dataset[(String, ScholixSummary)] = spark.read.load(summaryPath).as[ScholixSummary]

2
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/sx/graph/SparkCreateSummaryObject.scala
View File

@ -34,7 +34,7 @@ object SparkCreateSummaryObject {
implicit val summaryEncoder:Encoder[ScholixSummary] = Encoders.kryo[ScholixSummary]
val ds:Dataset[Result] = spark.read.load(s"$sourcePath/*").as[Oaf].filter(r => r.isInstanceOf[Result]).map(r => r.asInstanceOf[Result])
val ds:Dataset[Result] = spark.read.load(s"$sourcePath/*").as[Result].filter(r=>r.getDataInfo== null || r.getDataInfo.getDeletedbyinference== false)
ds.repartition(6000).map(r => ScholixUtils.resultToSummary(r)).filter(s => s!= null).write.mode(SaveMode.Overwrite).save(targetPath)