Skip to content

[SPARK-15764][SQL] Replace N^2 loop in BindReferences #13505

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
JoshRosen wants to merge 14 commits into from
Closed

[SPARK-15764][SQL] Replace N^2 loop in BindReferences #13505

Changes from 1 commit
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
6216e94
Replace N^2 loop in BindReferences.
JoshRosen Jun 4, 2016
b1a7646
Fix Java 7 compilation issue.
JoshRosen Jun 4, 2016
38e8a99
Whoops, getOrDefault is also new in Java 8.
JoshRosen Jun 4, 2016
0b412b0
Amortize map construction.
JoshRosen Jun 4, 2016
bc17587
Merge remote-tracking branch 'origin/master' into bind-references-imp…
JoshRosen Jun 4, 2016
e7c4150
Fix similar loop in InMemoryTableScanExec.
JoshRosen Jun 4, 2016
210dbd3
Rename getOrdinal to getOrdinalWithExprId
JoshRosen Jun 4, 2016
dd94e29
Delete some related yet dead code in AttributeMap.
JoshRosen Jun 4, 2016
b933fe0
Code style and readability improvements.
JoshRosen Jun 4, 2016
4efd3ee
Comments.
JoshRosen Jun 4, 2016
5504b6c
Address Eric's comments.
JoshRosen Jun 5, 2016
bdb68ad
Merge remote-tracking branch 'origin/master' into bind-references-imp…
JoshRosen Jun 6, 2016
99197b7
Add missing `@transient` annotations.
JoshRosen Jun 6, 2016
5e9c258
Also extend Serializable.
JoshRosen Jun 6, 2016
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
14 changes: 12 additions & 2 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/BoundAttribute.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -84,17 +84,27 @@ object BindReferences extends Logging {
expression: A,
input: Seq[Attribute],
Copy link
Contributor Author

@JoshRosen JoshRosen Jun 4, 2016

Choose a reason for hiding this comment

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

I wonder whether we can push the map construction up one level so that we can amortize its cost across multiple bindReference calls.

Copy link
Contributor Author

@JoshRosen JoshRosen Jun 4, 2016

Choose a reason for hiding this comment

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

Actually, yeah: in GenerateMutableProjection we use the same InputSchema for every expression.

Copy link
Contributor Author

@JoshRosen JoshRosen Jun 4, 2016

Choose a reason for hiding this comment

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

I think we should add an overload which takes a sequence of expressions and binds all of their references. We should then replace the call sites in the various projection operators.

allowFailures: Boolean = false): A = {
val inputArr = input.toArray
val inputToOrdinal = {
val map = new java.util.HashMap[ExprId, Int](inputArr.length * 2)
var index = 0
input.foreach { attr =>
map.putIfAbsent(attr.exprId, index)
index += 1
}
map
}
expression.transform { case a: AttributeReference =>
attachTree(a, "Binding attribute") {
val ordinal = input.indexWhere(_.exprId == a.exprId)
val ordinal = inputToOrdinal.getOrDefault(a.exprId, -1)
if (ordinal == -1) {
if (allowFailures) {
a
} else {
sys.error(s"Couldn't find $a in ${input.mkString("[", ",", "]")}")
}
} else {
BoundReference(ordinal, a.dataType, input(ordinal).nullable)
BoundReference(ordinal, a.dataType, inputArr(ordinal).nullable)
}
}
}.asInstanceOf[A] // Kind of a hack, but safe. TODO: Tighten return type when possible.
Expand Down