[SPARK-33456][SQL][TEST] Add end-to-end test for subexpression elimin…

…ation

### What changes were proposed in this pull request?

This patch proposes to add end-to-end test for subexpression elimination.

### Why are the changes needed?

We have subexpression elimination feature for expression evaluation but we don't have end-to-end tests for the feature. We should have one to make sure we don't break it.

### Does this PR introduce _any_ user-facing change?

No, dev only.

### How was this patch tested?

Unit tests.

Closes #30381 from viirya/SPARK-33456.

Authored-by: Liang-Chi Hsieh <viirya@gmail.com>
Signed-off-by: HyukjinKwon <gurwls223@apache.org>

sql/core/src/test/resources/sql-tests/inputs/subexp-elimination.sql

@@ -0,0 +1,37 @@
+-- Test for subexpression elimination.
+
+--SET spark.sql.optimizer.enableJsonExpressionOptimization=false
+
+--CONFIG_DIM1 spark.sql.codegen.wholeStage=true
+--CONFIG_DIM1 spark.sql.codegen.wholeStage=false
+
+--CONFIG_DIM2 spark.sql.codegen.factoryMode=CODEGEN_ONLY
+--CONFIG_DIM2 spark.sql.codegen.factoryMode=NO_CODEGEN
+
+--CONFIG_DIM3 SUBEXPRESSION_ELIMINATION_ENABLED=true
+--CONFIG_DIM3 SUBEXPRESSION_ELIMINATION_ENABLED=false
+
+-- Test data.
+CREATE OR REPLACE TEMPORARY VIEW testData AS SELECT * FROM VALUES
+('{"a":1, "b":"2"}', '[{"a": 1, "b":2}, {"a":2, "b":2}]'), ('{"a":1, "b":"2"}', null), ('{"a":2, "b":"3"}', '[{"a": 3, "b":4}, {"a":4, "b":5}]'), ('{"a":5, "b":"6"}', '[{"a": 6, "b":7}, {"a":8, "b":9}]'), (null, '[{"a": 1, "b":2}, {"a":2, "b":2}]')
+AS testData(a, b);
+
+SELECT from_json(a, 'struct<a:int,b:string>').a, from_json(a, 'struct<a:int,b:string>').b, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].b FROM testData;
+
+SELECT if(from_json(a, 'struct<a:int,b:string>').a > 1, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].a + 1) FROM testData;
+
+SELECT if(isnull(from_json(a, 'struct<a:int,b:string>').a), from_json(b, 'array<struct<a:int,b:int>>')[0].b + 1, from_json(b, 'array<struct<a:int,b:int>>')[0].b) FROM testData;
+
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(a, 'struct<a:int,b:string>').b when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(a, 'struct<a:int,b:string>').b + 1 else from_json(a, 'struct<a:int,b:string>').b + 2 end FROM testData;
+
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(b, 'array<struct<a:int,b:int>>')[0].b when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(b, 'array<struct<a:int,b:int>>')[0].b + 1 else from_json(b, 'array<struct<a:int,b:int>>')[0].b + 2 end FROM testData;
+
+-- With non-deterministic expressions.
+SELECT from_json(a, 'struct<a:int,b:string>').a + random() > 2, from_json(a, 'struct<a:int,b:string>').b, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].b + + random() > 2 FROM testData;
+
+SELECT if(from_json(a, 'struct<a:int,b:string>').a + random() > 5, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].a + 1) FROM testData;
+
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(a, 'struct<a:int,b:string>').b + random() > 5 when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(a, 'struct<a:int,b:string>').b + 1 + random() > 2 else from_json(a, 'struct<a:int,b:string>').b + 2 + random() > 5 end FROM testData;
+
+-- Clean up
+DROP VIEW IF EXISTS testData;

sql/core/src/test/resources/sql-tests/results/subexp-elimination.sql.out

@@ -0,0 +1,116 @@
+-- Automatically generated by SQLQueryTestSuite
+-- Number of queries: 10
+
+
+-- !query
+CREATE OR REPLACE TEMPORARY VIEW testData AS SELECT * FROM VALUES
+('{"a":1, "b":"2"}', '[{"a": 1, "b":2}, {"a":2, "b":2}]'), ('{"a":1, "b":"2"}', null), ('{"a":2, "b":"3"}', '[{"a": 3, "b":4}, {"a":4, "b":5}]'), ('{"a":5, "b":"6"}', '[{"a": 6, "b":7}, {"a":8, "b":9}]'), (null, '[{"a": 1, "b":2}, {"a":2, "b":2}]')
+AS testData(a, b)
+-- !query schema
+struct<>
+-- !query output
+
+
+
+-- !query
+SELECT from_json(a, 'struct<a:int,b:string>').a, from_json(a, 'struct<a:int,b:string>').b, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].b FROM testData
+-- !query schema
+struct<from_json(a).a:int,from_json(a).b:string,from_json(b)[0].a:int,from_json(b)[0].b:int>
+-- !query output
+1	2	1	2
+1	2	NULL	NULL
+2	3	3	4
+5	6	6	7
+NULL	NULL	1	2
+
+
+-- !query
+SELECT if(from_json(a, 'struct<a:int,b:string>').a > 1, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].a + 1) FROM testData
+-- !query schema
+struct<(IF((from_json(a).a > 1), from_json(b)[0].a, (from_json(b)[0].a + 1))):int>
+-- !query output
+2
+2
+3
+6
+NULL
+
+
+-- !query
+SELECT if(isnull(from_json(a, 'struct<a:int,b:string>').a), from_json(b, 'array<struct<a:int,b:int>>')[0].b + 1, from_json(b, 'array<struct<a:int,b:int>>')[0].b) FROM testData
+-- !query schema
+struct<(IF((from_json(a).a IS NULL), (from_json(b)[0].b + 1), from_json(b)[0].b)):int>
+-- !query output
+2
+3
+4
+7
+NULL
+
+
+-- !query
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(a, 'struct<a:int,b:string>').b when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(a, 'struct<a:int,b:string>').b + 1 else from_json(a, 'struct<a:int,b:string>').b + 2 end FROM testData
+-- !query schema
+struct<CASE WHEN (from_json(a).a > 5) THEN from_json(a).b WHEN (from_json(a).a > 4) THEN CAST((CAST(from_json(a).b AS DOUBLE) + CAST(1 AS DOUBLE)) AS STRING) ELSE CAST((CAST(from_json(a).b AS DOUBLE) + CAST(2 AS DOUBLE)) AS STRING) END:string>
+-- !query output
+4.0
+4.0
+5.0
+7.0
+NULL
+
+
+-- !query
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(b, 'array<struct<a:int,b:int>>')[0].b when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(b, 'array<struct<a:int,b:int>>')[0].b + 1 else from_json(b, 'array<struct<a:int,b:int>>')[0].b + 2 end FROM testData
+-- !query schema
+struct<CASE WHEN (from_json(a).a > 5) THEN from_json(b)[0].b WHEN (from_json(a).a > 4) THEN (from_json(b)[0].b + 1) ELSE (from_json(b)[0].b + 2) END:int>
+-- !query output
+4
+4
+6
+8
+NULL
+
+
+-- !query
+SELECT from_json(a, 'struct<a:int,b:string>').a + random() > 2, from_json(a, 'struct<a:int,b:string>').b, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].b + + random() > 2 FROM testData
+-- !query schema
+struct<((CAST(from_json(a).a AS DOUBLE) + rand()) > CAST(2 AS DOUBLE)):boolean,from_json(a).b:string,from_json(b)[0].a:int,((CAST(from_json(b)[0].b AS DOUBLE) + (+ rand())) > CAST(2 AS DOUBLE)):boolean>
+-- !query output
+NULL	NULL	1	true
+false	2	1	true
+false	2	NULL	NULL
+true	3	3	true
+true	6	6	true
+
+
+-- !query
+SELECT if(from_json(a, 'struct<a:int,b:string>').a + random() > 5, from_json(b, 'array<struct<a:int,b:int>>')[0].a, from_json(b, 'array<struct<a:int,b:int>>')[0].a + 1) FROM testData
+-- !query schema
+struct<(IF(((CAST(from_json(a).a AS DOUBLE) + rand()) > CAST(5 AS DOUBLE)), from_json(b)[0].a, (from_json(b)[0].a + 1))):int>
+-- !query output
+2
+2
+4
+6
+NULL
+
+
+-- !query
+SELECT case when from_json(a, 'struct<a:int,b:string>').a > 5 then from_json(a, 'struct<a:int,b:string>').b + random() > 5 when from_json(a, 'struct<a:int,b:string>').a > 4 then from_json(a, 'struct<a:int,b:string>').b + 1 + random() > 2 else from_json(a, 'struct<a:int,b:string>').b + 2 + random() > 5 end FROM testData
+-- !query schema
+struct<CASE WHEN (from_json(a).a > 5) THEN ((CAST(from_json(a).b AS DOUBLE) + rand()) > CAST(5 AS DOUBLE)) WHEN (from_json(a).a > 4) THEN (((CAST(from_json(a).b AS DOUBLE) + CAST(1 AS DOUBLE)) + rand()) > CAST(2 AS DOUBLE)) ELSE (((CAST(from_json(a).b AS DOUBLE) + CAST(2 AS DOUBLE)) + rand()) > CAST(5 AS DOUBLE)) END:boolean>
+-- !query output
+NULL
+false
+false
+true
+true
+
+
+-- !query
+DROP VIEW IF EXISTS testData
+-- !query schema
+struct<>
+-- !query output
+