[
https://issues.apache.org/jira/browse/SPARK-15810?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15369509#comment-15369509
]
Amit Sela edited comment on SPARK-15810 at 7/10/16 8:59 AM:
------------------------------------------------------------
Running the (sort of) same Java code:
{code}
SparkSession session = SparkSession.builder()
.appName("TestAggregatorJava")
.master("local[*]")
.getOrCreate();
Dataset<Tuple2<String, Integer>> ds1 = session.createDataset(Arrays.asList(
new Tuple2<>("a", 1),
new Tuple2<>("a", 2),
new Tuple2<>("a", 3)
), Encoders.tuple(Encoders.STRING(), Encoders.INT()));
Dataset<Tuple2<String, Integer>> ds2 = ds1.map(
new MapFunction<Tuple2<String, Integer>, Tuple2<String, Integer>>() {
@Override
public Tuple2<String, Integer> call(Tuple2<String, Integer> value) throws
Exception {
if (value._2() > 1) {
return value;
} else {
return new Tuple2<>(value._1, null);
}
}
}, Encoders.tuple(Encoders.STRING(), Encoders.INT()));
Dataset<Tuple2<String, Integer>> ds3 = ds2.groupByKey(
new MapFunction<Tuple2<String,Integer>, String>() {
@Override
public String call(Tuple2<String, Integer> value) throws Exception {
return value._1();
}
}, Encoders.STRING()).agg(new Aggregator<Tuple2<String, Integer>, Integer,
Integer>() {
@Override
public Integer zero() {
return null;
}
@Override
public Integer reduce(Integer b, Tuple2<String, Integer> a) {
return merge(b, a._2());
}
@Override
public Integer merge(Integer b1, Integer b2) {
if (b1 == null) {
return b2;
} else if (b2 == null){
return b1;
} else {
return b1 + b2;
}
}
@Override
public Integer finish(Integer reduction) {
return reduction;
}
@Override
public Encoder<Integer> bufferEncoder() {
return Encoders.INT();
}
@Override
public Encoder<Integer> outputEncoder() {
return Encoders.INT();
}
}.toColumn());
ds3.printSchema();
ds3.show();
}
{code}
I get this schema:
{noformat}
root
|-- value: string (nullable = true)
|-- (scala.Tuple2): integer (nullable = true)
{noformat}
And this result:
{noformat}
+-----+--------------+
|value|(scala.Tuple2)|
+-----+--------------+
| a| null|
+-----+--------------+
{noformat}
As for Scala, it's clear that `Option` is preferred on `null`, but because
Dataset API is supposed to support Java as well, it should not discard the
aggregation if the zero method returns null.
For Java, I currently use Guava's `Optional` but that just seems cumbersome to
me.
was (Author: amitsela):
Running the (sort of) same Java code:
{code}
SparkSession session = SparkSession.builder()
.appName("TestAggregatorJava")
.master("local[*]")
.getOrCreate();
Dataset<Tuple2<String, Integer>> ds1 = session.createDataset(Arrays.asList(
new Tuple2<>("a", 1),
new Tuple2<>("a", 2),
new Tuple2<>("a", 3)
), Encoders.tuple(Encoders.STRING(), Encoders.INT()));
Dataset<Tuple2<String, Integer>> ds2 = ds1.map(
new MapFunction<Tuple2<String, Integer>, Tuple2<String, Integer>>() {
@Override
public Tuple2<String, Integer> call(Tuple2<String, Integer> value) throws
Exception {
if (value._2() > 1) {
return value;
} else {
return new Tuple2<>(value._1, null);
}
}
}, Encoders.tuple(Encoders.STRING(), Encoders.INT()));
Dataset<Tuple2<String, Integer>> ds3 = ds2.groupByKey(
new MapFunction<Tuple2<String,Integer>, String>() {
@Override
public String call(Tuple2<String, Integer> value) throws Exception {
return value._1();
}
}, Encoders.STRING()).agg(new Aggregator<Tuple2<String, Integer>, Integer,
Integer>() {
@Override
public Integer zero() {
return null;
}
@Override
public Integer reduce(Integer b, Tuple2<String, Integer> a) {
return merge(b, a._2());
}
@Override
public Integer merge(Integer b1, Integer b2) {
if (b1 == null) {
return b2;
} else if (b2 == null){
return b1;
} else {
return b1 + b2;
}
}
@Override
public Integer finish(Integer reduction) {
return reduction;
}
@Override
public Encoder<Integer> bufferEncoder() {
return Encoders.INT();
}
@Override
public Encoder<Integer> outputEncoder() {
return Encoders.INT();
}
}.toColumn());
ds3.printSchema();
ds3.show();
}
{code}
I get this schema:
{noformat}
root
|-- value: string (nullable = true)
|-- (scala.Tuple2): integer (nullable = true)
{noformat}
And this result:
{noformat}
+-----+--------------+
|value|(scala.Tuple2)|
+-----+--------------+
| a| null|
+-----+--------------+
{noformat}
As for Scala, it's clear that {noformat}Option{noformat} is preferred on
`null`, but because Dataset API is supposed to support Java as well, it should
not discard the aggregation if the zero method returns null.
For Java, I currently use Guava's `Optional` but that just seems cumbersome to
me.
> Aggregator doesn't play nice with Option
> ----------------------------------------
>
> Key: SPARK-15810
> URL: https://issues.apache.org/jira/browse/SPARK-15810
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Environment: spark 2.0.0-SNAPSHOT
> Reporter: koert kuipers
>
> {code}
> val ds1 = List(("a", 1), ("a", 2), ("a", 3)).toDS
> val ds2 = ds1.map{ case (k, v) => (k, if (v > 1) Some(v) else None) }
> val ds3 = ds2.groupByKey(_._1).agg(new Aggregator[(String, Option[Int]),
> Option[Int], Option[Int]]{
> def zero: Option[Int] = None
> def reduce(b: Option[Int], a: (String, Option[Int])): Option[Int] =
> b.map(bv => a._2.map(av => bv + av).getOrElse(bv)).orElse(a._2)
> def merge(b1: Option[Int], b2: Option[Int]): Option[Int] = b1.map(b1v =>
> b2.map(b2v => b1v + b2v).getOrElse(b1v)).orElse(b2)
> def finish(reduction: Option[Int]): Option[Int] = reduction
> def bufferEncoder: Encoder[Option[Int]] = implicitly[Encoder[Option[Int]]]
> def outputEncoder: Encoder[Option[Int]] = implicitly[Encoder[Option[Int]]]
> }.toColumn)
> ds3.printSchema
> ds3.show
> {code}
> i get as output a somewhat odd looking schema, and after that the program
> just hangs pinning one cpu at 100%. the data never shows.
> output:
> {noformat}
> root
> |-- value: string (nullable = true)
> |-- $anon$1(scala.Tuple2): struct (nullable = true)
> | |-- value: integer (nullable = true)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
