Skip to content

[SPARK-31705][SQL] Push more possible predicates through Join via CNF conversion #28733

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
gengliangwang wants to merge 18 commits into from
Closed

[SPARK-31705][SQL] Push more possible predicates through Join via CNF conversion #28733

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
054594f
add rule PushCNFPredicateThroughJoin
gengliangwang Jun 5, 2020
6591680
From Yuming:[SPARK-31705][SQL] Push predicate through join by rewriti…
wangyum Jun 5, 2020
3497b3c
fix test failure
gengliangwang Jun 5, 2020
ada4135
Add test rewrite complex join predicates to conjunctive normal form
wangyum Jun 5, 2020
7e4b019
increase threshold and test case
gengliangwang Jun 5, 2020
0c92c1c
fix test case; reduce threshold default value
gengliangwang Jun 6, 2020
7853f67
address comments and add test cases
gengliangwang Jun 8, 2020
84e89de
revise
gengliangwang Jun 8, 2020
0af4c48
lazy val
gengliangwang Jun 9, 2020
95ee45e
update doc
gengliangwang Jun 9, 2020
6bf4747
remove assert
gengliangwang Jun 9, 2020
fa03b00
add back warning
gengliangwang Jun 9, 2020
c225f74
address comments
gengliangwang Jun 10, 2020
296068c
address comments
gengliangwang Jun 10, 2020
377f9d8
address comments
gengliangwang Jun 10, 2020
be79ab7
update test case
gengliangwang Jun 10, 2020
af018be
address comments
gengliangwang Jun 10, 2020
b42ce1d
fix build
gengliangwang Jun 11, 2020
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
Prev Previous commit
Next Next commit
fix test failure
  • Loading branch information
@gengliangwang
gengliangwang committed Jun 10, 2020
commit 3497b3c63848236274050a78b6d7bf19b314772d
2 changes: 1 addition & 1 deletion .../src/main/scala/org/apache/spark/sql/catalyst/optimizer/PushCNFPredicateThroughJoin.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -136,4 +136,4 @@ object PushCNFPredicateThroughJoin extends Rule[LogicalPlan] with PredicateHelpe
}
}
}
}
}
2 changes: 1 addition & 1 deletion sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -556,7 +556,7 @@ object SQLConf {
.intConf
.checkValue(_ >= 0,
"The depth of the maximum rewriting conjunction normal form must be positive.")
.createWithDefault(10)
.createWithDefault(20)

val ESCAPED_STRING_LITERALS = buildConf("spark.sql.parser.escapedStringLiterals")
.internal()
Expand Down
32 changes: 6 additions & 26 deletions ...catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ class FilterPushdownSuite extends PlanTest {
object Optimize extends RuleExecutor[LogicalPlan] {

override protected val blacklistedOnceBatches: Set[String] =
Set("Push predicate through join by CNF")
Set("Push CNF predicate through join")

val batches =
Batch("Subqueries", Once,
Expand All @@ -45,7 +45,7 @@ class FilterPushdownSuite extends PlanTest {
BooleanSimplification,
PushPredicateThroughJoin,
CollapseProject) ::
Batch("Push predicate through join by CNF", Once,
Batch("Push CNF predicate through join", Once,
PushCNFPredicateThroughJoin) :: Nil
}

Expand Down Expand Up @@ -1340,26 +1340,6 @@ class FilterPushdownSuite extends PlanTest {
comparePlans(optimized, correctAnswer)
}

test("inner join: rewrite to conjunctive normal form avoid generating too many predicates") {
val x = testRelation.subquery('x)
val y = testRelation.subquery('y)

val originalQuery = {
x.join(y, condition = Some(("x.b".attr === "y.b".attr)
&& ((("x.a".attr > 3) && ("x.a".attr < 13) && ("y.c".attr <= 5))
|| (("y.a".attr > 2) && ("y.c".attr < 1)))))
}

val optimized = Optimize.execute(originalQuery.analyze)
val left = testRelation.subquery('x)
val right = testRelation.where('c <= 5 || ('a > 2 && 'c < 1)).subquery('y)
val correctAnswer = left.join(right, condition = Some("x.b".attr === "y.b".attr
&& ((("x.a".attr > 3) && ("x.a".attr < 13) && ("y.c".attr <= 5))
|| (("y.a".attr > 2) && ("y.c".attr < 1))))).analyze

comparePlans(optimized, correctAnswer)
}

test(s"Disable rewrite to CNF by setting ${SQLConf.MAX_CNF_NODE_COUNT.key}=0") {
val x = testRelation.subquery('x)
val y = testRelation.subquery('y)
Expand All @@ -1370,14 +1350,14 @@ class FilterPushdownSuite extends PlanTest {
|| (("y.a".attr > 2) && ("y.c".attr < 1)))))
}

Seq(0, 10).foreach { depth =>
withSQLConf(SQLConf.MAX_CNF_NODE_COUNT.key -> depth.toString) {
Seq(0, 10).foreach { count =>
withSQLConf(SQLConf.MAX_CNF_NODE_COUNT.key -> count.toString) {
val optimized = Optimize.execute(originalQuery.analyze)
val (left, right) = if (depth == 0) {
val (left, right) = if (count == 0) {
(testRelation.subquery('x), testRelation.subquery('y))
} else {
(testRelation.subquery('x),
testRelation.where('c <= 5 || ('a > 2 && 'c < 1)).subquery('y))
testRelation.where(('c <= 5 || 'c < 1) && ('c <=5 || 'a > 2)).subquery('y))
Copy link
Member Author

@gengliangwang gengliangwang Jun 5, 2020

Choose a reason for hiding this comment

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

@wangyum To make it simple, this PR didn't convert the pushed down predicate to a shorter form.
We can have a follow-up PR if you like that feature.

}
val correctAnswer = left.join(right, condition = Some("x.b".attr === "y.b".attr
&& ((("x.a".attr > 3) && ("x.a".attr < 13) && ("y.c".attr <= 5))
Expand Down