Github user tejasapatil commented on a diff in the pull request:
https://github.com/apache/spark/pull/17993#discussion_r118852031
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/expressions.scala
---
@@ -54,6 +54,62 @@ object ConstantFolding extends Rule[LogicalPlan] {
}
}
+/**
+ * Substitutes [[Attribute Attributes]] which can be statically evaluated
with their corresponding
+ * value in conjunctive [[Expression Expressions]]
+ * eg.
+ * {{{
+ * SELECT * FROM table WHERE i = 5 AND j = i + 3
+ * ==> SELECT * FROM table WHERE i = 5 AND j = 8
+ * }}}
+ *
+ * Approach used:
+ * - Start from AND operator as the root
+ * - Get all the children conjunctive predicates which are EqualTo /
EqualNullSafe such that they
+ * don't have a `NOT` or `OR` operator in them
+ * - Populate a mapping of attribute => constant value by looking at all
the equals predicates
+ * - Using this mapping, replace occurrence of the attributes with the
corresponding constant values
+ * in the AND node.
+ */
+object ConstantPropagation extends Rule[LogicalPlan] with PredicateHelper {
+ private def containsNonConjunctionPredicates(expression: Expression):
Boolean = expression.find {
+ case _: Not | _: Or => true
+ case _ => false
+ }.isDefined
+
+ def apply(plan: LogicalPlan): LogicalPlan = plan transform {
+ case f: Filter => f transformExpressionsUp {
+ case and: And =>
+ val conjunctivePredicates =
+ splitConjunctivePredicates(and)
+ .filter(expr => expr.isInstanceOf[EqualTo] ||
expr.isInstanceOf[EqualNullSafe])
+ .filterNot(expr => containsNonConjunctionPredicates(expr))
+
+ val equalityPredicates = conjunctivePredicates.collect {
+ case e @ EqualTo(left: AttributeReference, right: Literal) =>
((left, right), e)
+ case e @ EqualTo(left: Literal, right: AttributeReference) =>
((right, left), e)
+ case e @ EqualNullSafe(left: AttributeReference, right: Literal)
=> ((left, right), e)
+ case e @ EqualNullSafe(left: Literal, right: AttributeReference)
=> ((right, left), e)
+ }
+
+ val constantsMap = AttributeMap(equalityPredicates.map(_._1))
+ val predicates = equalityPredicates.map(_._2).toSet
+
+ def replaceConstants(expression: Expression) = expression
transform {
+ case a: AttributeReference =>
+ constantsMap.get(a) match {
+ case Some(literal) => literal
+ case None => a
+ }
+ }
+
+ and transform {
+ case e @ EqualTo(_, _) if !predicates.contains(e) =>
replaceConstants(e)
--- End diff --
Here is the behavior with the PR. Seems reasonable because `a = 1` has to
be true so `(a = 1 or b = 2)` would always be `true` and can be eliminated.
```
scala> hc.sql(" select * from bla where (a = 1 or b = 2) and a = 1
").explain(true)
== Physical Plan ==
*Project [a#34, b#35]
+- *Filter (isnotnull(a#34) && (a#34 = 1))
+- *FileScan ....
```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
