zml1206 opened a new pull request, #46085:
URL: https://github.com/apache/spark/pull/46085
### What changes were proposed in this pull request?
Add rules `ConstantFolding`、`BooleanSimplification` and `PruneFilters` at
last of push extra predicate through join batch to optimize predicate.
For example:
```
spark.sql("select 1 as a, 2 as
b").write.mode("overwrite").parquet("tmp/t1")
spark.sql("select 1 as a").write.mode("overwrite").parquet("tmp/t2")
spark.sql("select 1 as a, 1 as
c").write.mode("overwrite").parquet("tmp/t3")
spark.read.parquet("tmp/t1").createOrReplaceTempView("t1")
spark.read.parquet("tmp/t2").createOrReplaceTempView("t2")
spark.read.parquet("tmp/t3").createOrReplaceTempView("t3")
spark.sql(
"""
|select t.*,t3.a as c from
|(
|select * from t1
|union all
|select *,1 as b from t2
|) t, t3
|where t.a=t3.a
|and (t.a > 1 or (t.b = 1 and t3.c=1))
|""".stripMargin)
```
before PR:
```
== Optimized Logical Plan ==
Project [a#0, b#1, a#6 AS c#11]
+- Join Inner, ((a#0 = a#6) AND ((a#0 > 1) OR ((b#1 = 1) AND (c#7 = 1))))
:- Union false, false
: :- Filter (isnotnull(a#0) AND ((a#0 > 1) OR (b#1 = 1)))
: : +- Relation [a#0,b#1] parquet
: +- Project [a#4, 1 AS b#10]
: +- Filter (isnotnull(a#4) AND ((a#4 > 1) OR (1 = 1)))
: +- Relation [a#4] parquet
+- Filter isnotnull(a#6)
+- Relation [a#6,c#7] parquet
```
after PR:
```
== Optimized Logical Plan ==
Project [a#15, b#16, a#21 AS c#26]
+- Join Inner, ((a#15 = a#21) AND ((a#15 > 1) OR ((b#16 = 1) AND (c#22 =
1))))
:- Union false, false
: :- Filter (isnotnull(a#15) AND ((a#15 > 1) OR (b#16 = 1)))
: : +- Relation [a#15,b#16] parquet
: +- Project [a#19, 1 AS b#25]
: +- Filter isnotnull(a#19)
: +- Relation [a#19] parquet
+- Filter isnotnull(a#21)
+- Relation [a#21,c#22] parquet
```
### Why are the changes needed?
Simplify plan and improve perfermance.
### Does this PR introduce _any_ user-facing change?
No.
### How was this patch tested?
Unit test.
### Was this patch authored or co-authored using generative AI tooling?
No.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
