miriam.baglioni
/
dnet-hadoop
forked from D-Net/dnet-hadoop
2
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity

reformat code

This commit is contained in:
Sandro La Bruzzo 2020-03-27 15:11:34 +01:00
parent a4b6a51168
commit 7cef698f36
2 changed files with 27 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
dhp-workflows/dhp-dedup-scholexplorer/src/main/java/eu/dnetlib/dedup/SparkCreateConnectedComponent.java
View File

@ -74,6 +74,7 @@ public class SparkCreateConnectedComponent {
}
public static long getHashcode(final String id) {
return Hashing.murmur3_128().hashUnencodedChars(id).asLong();
return Hashing.murmur3_128().hashString(id).asLong();
}
}

23
dhp-workflows/dhp-graph-provision-scholexplorer/src/main/java/eu/dnetlib/dhp/provision/DatasetJoiner.scala
View File

@ -8,22 +8,33 @@ object DatasetJoiner {
def startJoin(spark: SparkSession, relPath:String, targetPath:String) {
val relation = spark.read.load(relPath)
val relatedPublication = relation.where("target like '50%'").groupBy("source").agg(count("target").as("publication")).select(col("source"). alias("p_source"), col("publication"))
val relatedDataset = relation.where("target like '60%'").groupBy("source").agg(count("target").as("dataset")).select(col("source"). alias("d_source"), col("dataset"))
val relatedUnknown = relation.where("target like '70%'").groupBy("source").agg(count("target").as("unknown")).select(col("source"). alias("u_source"), col("unknown"))
val relatedPublication = relation
.where("target like '50%'")
.groupBy("source")
.agg(count("target").as("publication"))
.select(col("source"). alias("p_source"), col("publication"))
val relatedDataset = relation
.where("target like '60%'")
.groupBy("source")
.agg(count("target").as("dataset"))
.select(col("source"). alias("d_source"), col("dataset"))
val relatedUnknown = relation
.where("target like '70%'")
.groupBy("source")
.agg(count("target").as("unknown"))
.select(col("source"). alias("u_source"), col("unknown"))
val firstJoin = relatedPublication
.join(relatedDataset,col("p_source").equalTo(col("d_source")),"full")
.select(coalesce(col("p_source"), col("d_source")).alias("id"),
.select( coalesce( col("p_source"), col("d_source")).alias("id"),
col("publication"),
col("dataset"))
.join(relatedUnknown, col("u_source").equalTo(col("id")),"full")
.select(coalesce(col("u_source"), col("id")).alias("source"),
.select( coalesce(col("u_source"), col("id")).alias("source"),
coalesce(col("publication"),lit(0)).alias("relatedPublication"),
coalesce(col("dataset"),lit(0)).alias("relatedDataset"),
coalesce(col("unknown"),lit(0)).alias("relatedUnknown")
)
firstJoin.write.mode("overwrite").save(targetPath)
}
}