Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SPARK-20253][SQL] Remove unnecessary nullchecks of a return value from Spark runtime routines in generated Java code #17569

Closed
wants to merge 9 commits into from
Closed

[SPARK-20253][SQL] Remove unnecessary nullchecks of a return value from Spark runtime routines in generated Java code #17569

Show file tree
Hide file tree
Changes from 8 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
4482e1c
initial commit
kiszk Apr 7, 2017
ae5e232
fix scala style errors
kiszk Apr 7, 2017
fc6caac
address review comments
kiszk Apr 8, 2017
41c96ab
address review comments
kiszk Apr 8, 2017
a39803a
add test suites
kiszk Apr 8, 2017
895a3be
address review comment
kiszk Apr 8, 2017
3080ac2
fix test failure
kiszk Apr 8, 2017
510fb53
address review comment
kiszk Apr 8, 2017
10cf4be
adress review comment
kiszk Apr 9, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 16 additions & 11 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/ScalaReflection.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -251,19 +251,22 @@ object ScalaReflection extends ScalaReflection {
getPath :: Nil)

case t if t <:< localTypeOf[java.lang.String] =>
Invoke(getPath, "toString", ObjectType(classOf[String]))
Invoke(getPath, "toString", ObjectType(classOf[String]), returnNullable = false)

case t if t <:< localTypeOf[java.math.BigDecimal] =>
Invoke(getPath, "toJavaBigDecimal", ObjectType(classOf[java.math.BigDecimal]))
Invoke(getPath, "toJavaBigDecimal", ObjectType(classOf[java.math.BigDecimal]),
returnNullable = false)

case t if t <:< localTypeOf[BigDecimal] =>
Invoke(getPath, "toBigDecimal", ObjectType(classOf[BigDecimal]))
Invoke(getPath, "toBigDecimal", ObjectType(classOf[BigDecimal]), returnNullable = false)

case t if t <:< localTypeOf[java.math.BigInteger] =>
Invoke(getPath, "toJavaBigInteger", ObjectType(classOf[java.math.BigInteger]))
Invoke(getPath, "toJavaBigInteger", ObjectType(classOf[java.math.BigInteger]),
returnNullable = false)

case t if t <:< localTypeOf[scala.math.BigInt] =>
Invoke(getPath, "toScalaBigInt", ObjectType(classOf[scala.math.BigInt]))
Invoke(getPath, "toScalaBigInt", ObjectType(classOf[scala.math.BigInt]),
returnNullable = false)

case t if t <:< localTypeOf[Array[_]] =>
val TypeRef(_, _, Seq(elementType)) = t
Expand All @@ -284,7 +287,7 @@ object ScalaReflection extends ScalaReflection {
val arrayCls = arrayClassFor(elementType)

if (elementNullable) {
Invoke(arrayData, "array", arrayCls)
Invoke(arrayData, "array", arrayCls, returnNullable = false)
} else {
val primitiveMethod = elementType match {
case t if t <:< definitions.IntTpe => "toIntArray"
Expand All @@ -297,7 +300,7 @@ object ScalaReflection extends ScalaReflection {
case other => throw new IllegalStateException("expect primitive array element type " +
"but got " + other)
}
Invoke(arrayData, primitiveMethod, arrayCls)
Invoke(arrayData, primitiveMethod, arrayCls, returnNullable = false)
}

case t if t <:< localTypeOf[Seq[_]] =>
Expand Down Expand Up @@ -330,19 +333,21 @@ object ScalaReflection extends ScalaReflection {
Invoke(
MapObjects(
p => deserializerFor(keyType, Some(p), walkedTypePath),
Invoke(getPath, "keyArray", ArrayType(schemaFor(keyType).dataType)),
Invoke(getPath, "keyArray", ArrayType(schemaFor(keyType).dataType),
returnNullable = false),
schemaFor(keyType).dataType),
"array",
ObjectType(classOf[Array[Any]]))
ObjectType(classOf[Array[Any]]), returnNullable = false)

val valueData =
Invoke(
MapObjects(
p => deserializerFor(valueType, Some(p), walkedTypePath),
Invoke(getPath, "valueArray", ArrayType(schemaFor(valueType).dataType)),
Invoke(getPath, "valueArray", ArrayType(schemaFor(valueType).dataType),
returnNullable = false),
schemaFor(valueType).dataType),
"array",
ObjectType(classOf[Array[Any]]))
ObjectType(classOf[Array[Any]]), returnNullable = false)

StaticInvoke(
ArrayBasedMapData.getClass,
Expand Down
19 changes: 11 additions & 8 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/encoders/RowEncoder.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,7 +89,7 @@ object RowEncoder {
udtClass,
Nil,
dataType = ObjectType(udtClass), false)
Invoke(obj, "serialize", udt, inputObject :: Nil)
Invoke(obj, "serialize", udt, inputObject :: Nil, returnNullable = false)

case TimestampType =>
StaticInvoke(
Expand Down Expand Up @@ -136,16 +136,18 @@ object RowEncoder {
case t @ MapType(kt, vt, valueNullable) =>
val keys =
Invoke(
Invoke(inputObject, "keysIterator", ObjectType(classOf[scala.collection.Iterator[_]])),
Invoke(inputObject, "keysIterator", ObjectType(classOf[scala.collection.Iterator[_]]),
returnNullable = false),
"toSeq",
ObjectType(classOf[scala.collection.Seq[_]]))
ObjectType(classOf[scala.collection.Seq[_]]), returnNullable = false)
val convertedKeys = serializerFor(keys, ArrayType(kt, false))

val values =
Invoke(
Invoke(inputObject, "valuesIterator", ObjectType(classOf[scala.collection.Iterator[_]])),
Invoke(inputObject, "valuesIterator", ObjectType(classOf[scala.collection.Iterator[_]]),
returnNullable = false),
"toSeq",
ObjectType(classOf[scala.collection.Seq[_]]))
ObjectType(classOf[scala.collection.Seq[_]]), returnNullable = false)
val convertedValues = serializerFor(values, ArrayType(vt, valueNullable))

NewInstance(
Expand Down Expand Up @@ -262,17 +264,18 @@ object RowEncoder {
input :: Nil)

case _: DecimalType =>
Invoke(input, "toJavaBigDecimal", ObjectType(classOf[java.math.BigDecimal]))
Invoke(input, "toJavaBigDecimal", ObjectType(classOf[java.math.BigDecimal]),
returnNullable = false)

case StringType =>
Invoke(input, "toString", ObjectType(classOf[String]))
Invoke(input, "toString", ObjectType(classOf[String]), returnNullable = false)
Copy link
Contributor

@cloud-fan cloud-fan Apr 8, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we check how many places we set returnNullable to true? If it's only a few, we can change the default value of returnNullable to false.

Copy link
Member Author

@kiszk kiszk Apr 8, 2017 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here is statistics for 59 call sites of Invoke().

18: dataType is primitive type
21: returnNullable is true (no specification at call site, as default)
19: returnNullable is false
1: set a variable to `returnNullable

What do you think?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ok let's keep the default value unchanged


case ArrayType(et, nullable) =>
val arrayData =
Invoke(
MapObjects(deserializerFor(_), input, et),
"array",
ObjectType(classOf[Array[_]]))
ObjectType(classOf[Array[_]]), returnNullable = false)
StaticInvoke(
scala.collection.mutable.WrappedArray.getClass,
ObjectType(classOf[Seq[_]]),
Expand Down
28 changes: 14 additions & 14 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/objects/objects.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -225,25 +225,26 @@ case class Invoke(
getFuncResult(ev.value, s"${obj.value}.$functionName($argString)")
} else {
val funcResult = ctx.freshName("funcResult")
// If the function can return null, we do an extra check to make sure our null bit is still
// set correctly.
val postNullCheckAndAssign = if (!returnNullable) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

how about just assignResult?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

+1

s"${ev.value} = (${ctx.boxedType(javaType)}) $funcResult;"
} else {
s"""
if ($funcResult != null) {
${ev.value} = (${ctx.boxedType(javaType)}) $funcResult;
} else {
${ev.isNull} = true;
}
"""
}
s"""
Object $funcResult = null;
${getFuncResult(funcResult, s"${obj.value}.$functionName($argString)")}
if ($funcResult == null) {
${ev.isNull} = true;
} else {
${ev.value} = (${ctx.boxedType(javaType)}) $funcResult;
}
$postNullCheckAndAssign
"""
}

// If the function can return null, we do an extra check to make sure our null bit is still set
// correctly.
val postNullCheck = if (ctx.defaultValue(dataType) == "null") {
s"${ev.isNull} = ${ev.value} == null;"
} else {
""
}

val code = s"""
${obj.code}
boolean ${ev.isNull} = true;
Expand All @@ -254,7 +255,6 @@ case class Invoke(
if (!${ev.isNull}) {
$evaluate
}
$postNullCheck
}
"""
ev.copy(code = code)
Expand Down
10 changes: 10 additions & 0 deletions sql/core/src/test/scala/org/apache/spark/sql/DatasetPrimitiveSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,6 +96,16 @@ class DatasetPrimitiveSuite extends QueryTest with SharedSQLContext {
checkDataset(dsBoolean.map(e => !e), false, true)
}

test("mapPrimitiveArray") {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

do these tests fail before this PR?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, I have just added to confirm this check works well.

val dsInt = Seq(Array(1, 2), Array(3, 4)).toDS()
checkDataset(dsInt.map(e => e), Array(1, 2), Array(3, 4))
checkDataset(dsInt.map(e => null: Array[Int]), null, null)

val dsDouble = Seq(Array(1D, 2D), Array(3D, 4D)).toDS()
checkDataset(dsDouble.map(e => e), Array(1D, 2D), Array(3D, 4D))
checkDataset(dsDouble.map(e => null: Array[Double]), null, null)
}

test("filter") {
val ds = Seq(1, 2, 3, 4).toDS()
checkDataset(
Expand Down