Handles EXPLAIN output diffs in PG15, Hash Agg/Join leverage

To handle differences in usage of GroupAggregate vs HashAggregate
or Merge Join vs Hash join in cases where this detail doesn't
seem to matter, we use coordinator_plan().
- coordinator_plan() is updated to remove "Result" lines

There are some cases where we have subplans so we add a new
function that prints all Task Count lines as well
- coordinator_plan_with_subplans()

Still not sure of the relevant PG commit
Could be db0d67db2401eb6238ccc04c6407a4fd4f985832
but disabling enable_group_by_reordering didn't help.

src/test/regress/expected/cte_inline.out

@@ -423,6 +423,8 @@ DEBUG:  join prunable for intervals [1073741824,2147483647] and [0,1073741823]
 (1 row)
 
 -- EXPLAIN should show the differences between MATERIALIZED and NOT MATERIALIZED
+\set VERBOSITY terse
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte_1 AS (SELECT * FROM test_table)
 SELECT
 	count(*)
@@ -431,36 +433,22 @@ FROM
 		JOIN
 	cte_1 as second_entry
 		USING (key);
+$Q$);
 DEBUG:  Router planner cannot handle multi-shard select queries
 DEBUG:  generating subplan XXX_1 for CTE cte_1: SELECT key, value, other_value FROM cte_inline.test_table
 DEBUG:  Router planner cannot handle multi-shard select queries
 DEBUG:  Plan XXX query after replacing subqueries and CTEs: SELECT count(*) AS count FROM ((SELECT intermediate_result.key, intermediate_result.value, intermediate_result.other_value FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(key integer, value text, other_value jsonb)) first_entry JOIN (SELECT intermediate_result.key, intermediate_result.value, intermediate_result.other_value FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(key integer, value text, other_value jsonb)) second_entry USING (key))
 DEBUG:  Creating router plan
-                                          QUERY PLAN
+      coordinator_plan_with_subplans
 ---------------------------------------------------------------------
  Custom Scan (Citus Adaptive)
    ->  Distributed Subplan XXX_1
          ->  Custom Scan (Citus Adaptive)
                Task Count: 4
-               Tasks Shown: One of 4
-               ->  Task
-                     Node: host=localhost port=xxxxx dbname=regression
-                     ->  Seq Scan on test_table_1960000 test_table
    Task Count: 1
-   Tasks Shown: All
-   ->  Task
-         Node: host=localhost port=xxxxx dbname=regression
-         ->  Aggregate
-               ->  Merge Join
-                     Merge Cond: (intermediate_result.key = intermediate_result_1.key)
-                     ->  Sort
-                           Sort Key: intermediate_result.key
-                           ->  Function Scan on read_intermediate_result intermediate_result
-                     ->  Sort
-                           Sort Key: intermediate_result_1.key
-                           ->  Function Scan on read_intermediate_result intermediate_result_1
-(21 rows)
+(5 rows)
 
+\set VERBOSITY default
 EXPLAIN (COSTS OFF) WITH cte_1 AS NOT MATERIALIZED (SELECT * FROM test_table)
 SELECT
 	count(*)

src/test/regress/expected/insert_select_repartition.out

@@ -702,23 +702,16 @@ PREPARE insert_plan AS
 INSERT INTO target_table
   SELECT a, max(b) FROM source_table
   WHERE a BETWEEN 1 AND 2 GROUP BY a;
+SELECT public.coordinator_plan($Q$
 EXPLAIN EXECUTE insert_plan;
-                                                   QUERY PLAN
+$Q$);
+                             coordinator_plan
 ---------------------------------------------------------------------
  Custom Scan (Citus INSERT ... SELECT)  (cost=0.00..0.00 rows=0 width=0)
    INSERT/SELECT method: repartition
    ->  Custom Scan (Citus Adaptive)  (cost=0.00..0.00 rows=100000 width=8)
          Task Count: 4
-         Tasks Shown: One of 4
-         ->  Task
-               Node: host=localhost port=xxxxx dbname=regression
-               ->  GroupAggregate  (cost=44.09..44.28 rows=11 width=8)
-                     Group Key: a
-                     ->  Sort  (cost=44.09..44.12 rows=11 width=8)
-                           Sort Key: a
-                           ->  Seq Scan on source_table_4213606 source_table  (cost=0.00..43.90 rows=11 width=8)
-                                 Filter: ((a >= 1) AND (a <= 2))
-(13 rows)
+(4 rows)
 
 SET client_min_messages TO DEBUG1;
 EXECUTE insert_plan;

src/test/regress/expected/multi_test_helpers.out

@@ -17,10 +17,15 @@ BEGIN
 END;
 $$LANGUAGE plpgsql;
 -- Create a function to ignore worker plans in explain output
+-- Also remove extra "->  Result" lines for PG15 support
 CREATE OR REPLACE FUNCTION coordinator_plan(explain_command text, out query_plan text)
 RETURNS SETOF TEXT AS $$
 BEGIN
   FOR query_plan IN execute explain_command LOOP
+    IF (query_plan LIKE '%->  Result%' OR query_plan = 'Result')
+    THEN
+        CONTINUE;
+    END IF;
     RETURN next;
     IF query_plan LIKE '%Task Count:%'
     THEN
@@ -29,6 +34,31 @@ BEGIN
   END LOOP;
   RETURN;
 END; $$ language plpgsql;
+-- Create a function to ignore worker plans in explain output
+-- It also shows task count for plan and subplans
+-- Also remove extra "->  Result" lines for PG15 support
+CREATE OR REPLACE FUNCTION coordinator_plan_with_subplans(explain_command text, out query_plan text)
+RETURNS SETOF TEXT AS $$
+DECLARE
+    task_count_line_reached boolean := false;
+BEGIN
+  FOR query_plan IN execute explain_command LOOP
+    IF (query_plan LIKE '%->  Result%' OR query_plan = 'Result') THEN
+        CONTINUE;
+    END IF;
+    IF NOT task_count_line_reached THEN
+        RETURN next;
+    END IF;
+    IF query_plan LIKE '%Task Count:%' THEN
+        IF NOT task_count_line_reached THEN
+            SELECT true INTO task_count_line_reached;
+        ELSE
+            RETURN next;
+        END IF;
+    END IF;
+  END LOOP;
+  RETURN;
+END; $$ language plpgsql;
 -- Create a function to ignore "->  Result" lines for PG15 support
 -- In PG15 there are some extra "->  Result" lines
 CREATE OR REPLACE FUNCTION plan_without_result_lines(explain_command text, out query_plan text)

src/test/regress/expected/multi_view.out

@@ -785,6 +785,7 @@ EXPLAIN (COSTS FALSE) SELECT user_id FROM recent_selected_users GROUP BY 1 ORDER
                                  Filter: ((value_1 >= 1) AND (value_1 < 3))
 (19 rows)
 
+SELECT public.coordinator_plan($Q$
 EXPLAIN (COSTS FALSE) SELECT *
 	FROM (
 		(SELECT user_id FROM recent_users)
@@ -792,32 +793,14 @@ EXPLAIN (COSTS FALSE) SELECT *
 		(SELECT user_id FROM selected_users) ) u
 	WHERE user_id < 4 AND user_id > 1
 	ORDER BY user_id;
-                                                                   QUERY PLAN
+$Q$);
+          coordinator_plan
 ---------------------------------------------------------------------
  Sort
    Sort Key: remote_scan.user_id
    ->  Custom Scan (Citus Adaptive)
          Task Count: 4
-         Tasks Shown: One of 4
-         ->  Task
-               Node: host=localhost port=xxxxx dbname=regression
-               ->  Unique
-                     ->  Sort
-                           Sort Key: recent_users.user_id
-                           ->  Append
-                                 ->  Subquery Scan on recent_users
-                                       ->  Sort
-                                             Sort Key: (max(users_table."time")) DESC
-                                             ->  GroupAggregate
-                                                   Group Key: users_table.user_id
-                                                   Filter: (max(users_table."time") > '2017-11-23 16:20:33.264457'::timestamp without time zone)
-                                                   ->  Sort
-                                                         Sort Key: users_table.user_id
-                                                         ->  Seq Scan on users_table_1400256 users_table
-                                                               Filter: ((user_id < 4) AND (user_id > 1))
-                                 ->  Seq Scan on users_table_1400256 users_table_1
-                                       Filter: ((value_1 >= 1) AND (value_1 < 3) AND (user_id < 4) AND (user_id > 1))
-(23 rows)
+(4 rows)
 
 EXPLAIN (COSTS FALSE) SELECT et.* FROM recent_10_users JOIN events_table et USING(user_id) ORDER BY et.time DESC LIMIT 10;
                                                   QUERY PLAN

src/test/regress/expected/subquery_view.out

@@ -578,11 +578,13 @@ SELECT create_reference_table('reference_table');
 
 (1 row)
 
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte AS (
   SELECT application_name AS text_col
   FROM pg_stat_activity
 ) SELECT * FROM reference_table JOIN cte USING (text_col);
-                                            QUERY PLAN
+$Q$);
+           coordinator_plan_with_subplans
 ---------------------------------------------------------------------
  Custom Scan (Citus Adaptive)
    ->  Distributed Subplan XXX_1
@@ -590,38 +592,17 @@ EXPLAIN (COSTS OFF) WITH cte AS (
    ->  Distributed Subplan XXX_2
          ->  Custom Scan (Citus Adaptive)
                Task Count: 1
-               Tasks Shown: All
-               ->  Task
-                     Node: host=localhost port=xxxxx dbname=regression
-                     ->  Hash Left Join
-                           Hash Cond: (intermediate_result.usesysid = u.oid)
-                           ->  Hash Left Join
-                                 Hash Cond: (intermediate_result.datid = d.oid)
-                                 ->  Function Scan on read_intermediate_result intermediate_result
-                                 ->  Hash
-                                       ->  Seq Scan on pg_database d
-                           ->  Hash
-                                 ->  Seq Scan on pg_authid u
    Task Count: 1
-   Tasks Shown: All
-   ->  Task
-         Node: host=localhost port=xxxxx dbname=regression
-         ->  Merge Join
-               Merge Cond: (intermediate_result.application_name = reference_table.text_col)
-               ->  Sort
-                     Sort Key: intermediate_result.application_name
-                     ->  Function Scan on read_intermediate_result intermediate_result
-               ->  Sort
-                     Sort Key: reference_table.text_col
-                     ->  Seq Scan on reference_table_1512000 reference_table
-(30 rows)
+(7 rows)
 
 CREATE OR REPLACE VIEW view_on_views AS SELECT pg_stat_activity.application_name, pg_locks.pid FROM pg_stat_activity, pg_locks;
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte AS (
   SELECT application_name AS text_col
   FROM view_on_views
 ) SELECT * FROM reference_table JOIN cte USING (text_col);
-                                      QUERY PLAN
+$Q$);
+              coordinator_plan_with_subplans
 ---------------------------------------------------------------------
  Custom Scan (Citus Adaptive)
    ->  Distributed Subplan XXX_1
@@ -629,18 +610,7 @@ EXPLAIN (COSTS OFF) WITH cte AS (
                ->  Function Scan on pg_stat_get_activity s
                ->  Function Scan on pg_lock_status l
    Task Count: 1
-   Tasks Shown: All
-   ->  Task
-         Node: host=localhost port=xxxxx dbname=regression
-         ->  Merge Join
-               Merge Cond: (intermediate_result.text_col = reference_table.text_col)
-               ->  Sort
-                     Sort Key: intermediate_result.text_col
-                     ->  Function Scan on read_intermediate_result intermediate_result
-               ->  Sort
-                     Sort Key: reference_table.text_col
-                     ->  Seq Scan on reference_table_1512000 reference_table
-(17 rows)
+(6 rows)
 
 DROP SCHEMA subquery_view CASCADE;
 NOTICE:  drop cascades to 19 other objects

src/test/regress/sql/cte_inline.sql

@@ -220,6 +220,10 @@ FROM
 		USING (key);
 
 -- EXPLAIN should show the differences between MATERIALIZED and NOT MATERIALIZED
+
+\set VERBOSITY terse
+
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte_1 AS (SELECT * FROM test_table)
 SELECT
 	count(*)
@@ -228,6 +232,9 @@ FROM
 		JOIN
 	cte_1 as second_entry
 		USING (key);
+$Q$);
+
+\set VERBOSITY default
 
 EXPLAIN (COSTS OFF) WITH cte_1 AS NOT MATERIALIZED (SELECT * FROM test_table)
 SELECT

src/test/regress/sql/insert_select_repartition.sql

@@ -354,7 +354,9 @@ INSERT INTO target_table
   SELECT a, max(b) FROM source_table
   WHERE a BETWEEN 1 AND 2 GROUP BY a;
 
+SELECT public.coordinator_plan($Q$
 EXPLAIN EXECUTE insert_plan;
+$Q$);
 
 SET client_min_messages TO DEBUG1;
 EXECUTE insert_plan;

src/test/regress/sql/multi_test_helpers.sql

@@ -20,10 +20,15 @@ END;
 $$LANGUAGE plpgsql;
 
 -- Create a function to ignore worker plans in explain output
+-- Also remove extra "->  Result" lines for PG15 support
 CREATE OR REPLACE FUNCTION coordinator_plan(explain_command text, out query_plan text)
 RETURNS SETOF TEXT AS $$
 BEGIN
   FOR query_plan IN execute explain_command LOOP
+    IF (query_plan LIKE '%->  Result%' OR query_plan = 'Result')
+    THEN
+        CONTINUE;
+    END IF;
     RETURN next;
     IF query_plan LIKE '%Task Count:%'
     THEN
@@ -33,6 +38,32 @@ BEGIN
   RETURN;
 END; $$ language plpgsql;
 
+-- Create a function to ignore worker plans in explain output
+-- It also shows task count for plan and subplans
+-- Also remove extra "->  Result" lines for PG15 support
+CREATE OR REPLACE FUNCTION coordinator_plan_with_subplans(explain_command text, out query_plan text)
+RETURNS SETOF TEXT AS $$
+DECLARE
+    task_count_line_reached boolean := false;
+BEGIN
+  FOR query_plan IN execute explain_command LOOP
+    IF (query_plan LIKE '%->  Result%' OR query_plan = 'Result') THEN
+        CONTINUE;
+    END IF;
+    IF NOT task_count_line_reached THEN
+        RETURN next;
+    END IF;
+    IF query_plan LIKE '%Task Count:%' THEN
+        IF NOT task_count_line_reached THEN
+            SELECT true INTO task_count_line_reached;
+        ELSE
+            RETURN next;
+        END IF;
+    END IF;
+  END LOOP;
+  RETURN;
+END; $$ language plpgsql;
+
 -- Create a function to ignore "->  Result" lines for PG15 support
 -- In PG15 there are some extra "->  Result" lines
 CREATE OR REPLACE FUNCTION plan_without_result_lines(explain_command text, out query_plan text)

src/test/regress/sql/multi_view.sql

@@ -374,6 +374,7 @@ VACUUM ANALYZE users_table;
 -- explain tests
 EXPLAIN (COSTS FALSE) SELECT user_id FROM recent_selected_users GROUP BY 1 ORDER BY 1;
 
+SELECT public.coordinator_plan($Q$
 EXPLAIN (COSTS FALSE) SELECT *
 	FROM (
 		(SELECT user_id FROM recent_users)
@@ -381,6 +382,7 @@ EXPLAIN (COSTS FALSE) SELECT *
 		(SELECT user_id FROM selected_users) ) u
 	WHERE user_id < 4 AND user_id > 1
 	ORDER BY user_id;
+$Q$);
 
 EXPLAIN (COSTS FALSE) SELECT et.* FROM recent_10_users JOIN events_table et USING(user_id) ORDER BY et.time DESC LIMIT 10;
 SET citus.subquery_pushdown to ON;

src/test/regress/sql/subquery_view.sql

@@ -427,17 +427,21 @@ SET client_min_messages TO DEFAULT;
 CREATE TABLE reference_table (text_col text, int_col int);
 SELECT create_reference_table('reference_table');
 
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte AS (
   SELECT application_name AS text_col
   FROM pg_stat_activity
 ) SELECT * FROM reference_table JOIN cte USING (text_col);
+$Q$);
 
 CREATE OR REPLACE VIEW view_on_views AS SELECT pg_stat_activity.application_name, pg_locks.pid FROM pg_stat_activity, pg_locks;
 
+SELECT public.coordinator_plan_with_subplans($Q$
 EXPLAIN (COSTS OFF) WITH cte AS (
   SELECT application_name AS text_col
   FROM view_on_views
 ) SELECT * FROM reference_table JOIN cte USING (text_col);
+$Q$);
 
 DROP SCHEMA subquery_view CASCADE;
 SET search_path TO public;