Skip to content

SPARK-3711: Optimize where in clause filter queries #2561

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
saucam wants to merge 6 commits into from
Closed

SPARK-3711: Optimize where in clause filter queries #2561

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
bee98aa
SPARK-3711: Optimize where in clause filter queries
Sep 28, 2014
430f5d1
SPARK-3711: Optimize the filter list in case of negative values as well
Oct 1, 2014
bd84c67
SPARK-3711: Incorporate review comments. Move optimization of In clau…
Oct 2, 2014
0fc902f
SPARK-3711: UnaryMinus will be handled by constantFolding
Oct 3, 2014
afedbcd
SPARK-3711: 1. Add test cases for InSet class in ExpressionEvaluation…
Oct 3, 2014
4bf2d19
SPARK-3711: 1. Fix code style and import order
Oct 6, 2014
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
17 changes: 16 additions & 1 deletion sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/predicates.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,7 @@ package org.apache.spark.sql.catalyst.expressions
import org.apache.spark.sql.catalyst.analysis.UnresolvedException
import org.apache.spark.sql.catalyst.plans.logical.LogicalPlan
import org.apache.spark.sql.catalyst.types.BooleanType

import scala.collection.immutable.HashSet

@marmbrus marmbrus Oct 6, 2014

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Scala import should go first. See here for more details.


object InterpretedPredicate {
def apply(expression: Expression, inputSchema: Seq[Attribute]): (Row => Boolean) =
Expand Down Expand Up @@ -95,6 +95,21 @@ case class In(value: Expression, list: Seq[Expression]) extends Predicate {
}
}

/**
* Evaluates to `true` if `list` contains `value`.

@marmbrus marmbrus Oct 2, 2014

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Add a comment that this is an optimized version of In for when all values of the inList are static.

*/
case class InSet(value: Expression, hset: HashSet[Any], child: Seq[Expression])
extends Predicate {

@marmbrus marmbrus Oct 6, 2014

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

This should actually only be indented 2 spaces (and I'd include a blank line after). Only wrapped arguments are indented 4.

def children = child

def nullable = true // TODO: Figure out correct nullability semantics of IN.
override def toString = s"$value IN ${hset.mkString("(", ",", ")")}"

@marmbrus marmbrus Oct 6, 2014

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

I'd actually say InSet here as the toString is mostly for debugging / understanding how queries will execute.


override def eval(input: Row): Any = {
hset.contains(value.eval(input))
}
}

case class And(left: Expression, right: Expression) extends BinaryPredicate {
def symbol = "&&"

Expand Down
12 changes: 11 additions & 1 deletion sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveQl.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,7 @@ import org.apache.spark.sql.catalyst.plans.logical
import org.apache.spark.sql.catalyst.plans.logical._
import org.apache.spark.sql.catalyst.types._

import scala.collection.immutable.HashSet
/* Implicit conversions */
import scala.collection.JavaConversions._

Expand Down Expand Up @@ -977,7 +978,16 @@ private[hive] object HiveQl {
case Token("TOK_FUNCTION", Token("TOK_ISNULL", Nil) :: child :: Nil) =>
IsNull(nodeToExpr(child))
case Token("TOK_FUNCTION", Token(IN(), Nil) :: value :: list) =>
In(nodeToExpr(value), list.map(nodeToExpr))
val valExpr = nodeToExpr(value)

@marmbrus marmbrus Oct 2, 2014

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Instead of hardwiring this into the parser, lets create a rule in Optimizer.scala that recognizes this pattern. That will allow this optimization to be more general. For example it could then easily apply to queries like ... WHERE x IN (1 + 1, 1 + 2, ...) since it'll happen along with ConstantFolding.

val listMap = list.map(nodeToExpr)
if (listMap.exists(e => ((!e.isInstanceOf[Literal]) &&
(!e.isInstanceOf[UnaryMinus]))) == false){
val hSet = listMap.map(e => e.eval(null))
InSet(valExpr, HashSet() ++ hSet, valExpr +: listMap)
}
else{
In(valExpr, listMap)
}
case Token("TOK_FUNCTION",
Token(BETWEEN(), Nil) ::
Token("KW_FALSE", Nil) ::
Expand Down