Skip to content

[SPARK-9401][SQL]Fully implement code generation for ConcatWs #8353

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

Closed
yjshen wants to merge 3 commits into from
Closed

[SPARK-9401][SQL]Fully implement code generation for ConcatWs #8353

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
affd497
codegen concatws
yjshen Aug 21, 2015
84c1dfb
address comments
yjshen Aug 22, 2015
eb29b30
address comments
yjshen Aug 22, 2015
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
42 changes: 39 additions & 3 deletions ...catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/stringExpressions.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,7 +72,7 @@ case class Concat(children: Seq[Expression]) extends Expression with ImplicitCas
* Returns null if the separator is null. Otherwise, concat_ws skips all null values.
*/
case class ConcatWs(children: Seq[Expression])
extends Expression with ImplicitCastInputTypes with CodegenFallback {
extends Expression with ImplicitCastInputTypes {

require(children.nonEmpty, s"$prettyName requires at least one argument.")

Expand Down Expand Up @@ -114,8 +114,44 @@ case class ConcatWs(children: Seq[Expression])
boolean ${ev.isNull} = ${ev.primitive} == null;
"""
} else {
// Contains a mix of strings and array<string>s. Fall back to interpreted mode for now.
super.genCode(ctx, ev)
val array = ctx.freshName("array")
val varargNum = ctx.freshName("varargNum")
val idxInVararg = ctx.freshName("idxInVararg")

val evals = children.map(_.gen(ctx))
val (varargCount, varargBuild) = children.tail.zip(evals.tail).map { case (child, eval) =>
child.dataType match {
case StringType =>
(s"$varargNum ++;",
s"$array[$idxInVararg ++] = ${eval.isNull} ? (UTF8String) null : ${eval.primitive};")
case _: ArrayType =>
val size = ctx.freshName("n")
(s"""
if (!${eval.isNull}) {
$varargNum += ${eval.primitive}.numElements();
}
""",
s"""
if (!${eval.isNull}) {
final int $size = ${eval.primitive}.numElements();
for (int j = 0; j < $size; j ++) {
$array[$idxInVararg ++] = ${ctx.getValue(eval.primitive, StringType, "j")};
}
}
""")
}
}.unzip

evals.map(_.code).mkString("\n") +
s"""
int $varargNum = 0;
Copy link
Contributor

@davies davies Aug 22, 2015

Choose a reason for hiding this comment

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

This could be the number of vargs that is UTF8String (${children.count(_ == StringType) - 1} ).

Copy link
Member Author

@yjshen yjshen Aug 22, 2015

Choose a reason for hiding this comment

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

I use this as the total num of strings, stringType + arrayType's element num.

Copy link
Contributor

@davies davies Aug 22, 2015

Choose a reason for hiding this comment

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

I means we can get rid of some many num ++

Copy link
Member Author

@yjshen yjshen Aug 22, 2015

Choose a reason for hiding this comment

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

ah, yes, I get this.

Copy link
Member Author

@yjshen yjshen Aug 22, 2015

Choose a reason for hiding this comment

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

@davies , more comments on this PR?

int $idxInVararg = 0;
${varargCount.mkString("\n")}
UTF8String[] $array = new UTF8String[$varargNum];
${varargBuild.mkString("\n")}
UTF8String ${ev.primitive} = UTF8String.concatWs(${evals.head.primitive}, $array);
boolean ${ev.isNull} = ${ev.primitive} == null;
"""
}
}
}
Expand Down