gianm commented on a change in pull request #9648: SQL: More straightforward
handling of join planning.
URL: https://github.com/apache/druid/pull/9648#discussion_r406461945
##########
File path:
sql/src/main/java/org/apache/druid/sql/calcite/rule/DruidJoinRule.java
##########
@@ -148,4 +246,148 @@ static boolean canHandleCondition(final RexNode
condition, final RelDataType lef
return retVal;
}
+
+ private static boolean isLeftExpression(final RexNode rexNode, final int
numLeftFields)
+ {
+ return
ImmutableBitSet.range(numLeftFields).contains(RelOptUtil.InputFinder.bits(rexNode));
+ }
+
+ private static boolean isRightInputRef(final RexNode rexNode, final int
numLeftFields)
+ {
+ return rexNode.isA(SqlKind.INPUT_REF) && ((RexInputRef)
rexNode).getIndex() >= numLeftFields;
+ }
+
+ @VisibleForTesting
+ static class ConditionAnalysis
+ {
+ /**
+ * Number of fields on the left-hand side. Useful for identifying if a
particular field is from on the left
+ * or right side of a join.
+ */
+ private final int numLeftFields;
+
+ /**
+ * Each equality subcondition is an equality of the form f(LeftRel) =
g(RightRel).
+ */
+ private final List<Pair<RexNode, RexInputRef>> equalitySubConditions;
+
+ /**
+ * Each literal subcondition is... a literal.
+ */
+ private final List<RexLiteral> literalSubConditions;
+
+ ConditionAnalysis(
+ int numLeftFields,
+ List<Pair<RexNode, RexInputRef>> equalitySubConditions,
+ List<RexLiteral> literalSubConditions
+ )
+ {
+ this.numLeftFields = numLeftFields;
+ this.equalitySubConditions = equalitySubConditions;
+ this.literalSubConditions = literalSubConditions;
+ }
+
+ public ConditionAnalysis pushThroughLeftProject(final Project leftProject)
+ {
+ // Pushing through the project will shift right-hand field references by
this amount.
+ final int rhsShift =
+ leftProject.getInput().getRowType().getFieldCount() -
leftProject.getRowType().getFieldCount();
+
+ return new ConditionAnalysis(
+ leftProject.getInput().getRowType().getFieldCount(),
+ equalitySubConditions
+ .stream()
Review comment:
Personally my strategy has been to use whatever seems most readable for code
that isn't performance critical (small collections, not called tons of times).
And for situations where performance is critical then to benchmark.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
