From 628bb5ca7f563ab7f11c373572145df403de6fef Mon Sep 17 00:00:00 2001
From: Matei Zaharia <matei@eecs.berkeley.edu>
Date: Thu, 12 Jul 2012 18:36:02 -0700
Subject: [PATCH] Allow null keys in Spark's reduce and group by
---
core/src/main/scala/spark/Partitioner.scala | 14 +++++++++-----
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/core/src/main/scala/spark/Partitioner.scala b/core/src/main/scala/spark/Partitioner.scala
index 0e45ebd35c..475da17de5 100644
--- a/core/src/main/scala/spark/Partitioner.scala
+++ b/core/src/main/scala/spark/Partitioner.scala
@@ -8,12 +8,16 @@ abstract class Partitioner extends Serializable {
class HashPartitioner(partitions: Int) extends Partitioner {
def numPartitions = partitions
- def getPartition(key: Any) = {
- val mod = key.hashCode % partitions
- if (mod < 0) {
- mod + partitions
+ def getPartition(key: Any): Int = {
+ if (key == null) {
+ return 0
} else {
- mod // Guard against negative hash codes
+ val mod = key.hashCode % partitions
+ if (mod < 0) {
+ mod + partitions
+ } else {
+ mod // Guard against negative hash codes
+ }
}
}
--
GitLab