-- ===================================================================
-- test recursive planning functionality with subqueries and CTEs
-- ===================================================================
CREATE SCHEMA subquery_and_ctes;
SET search_path TO subquery_and_ctes, public;
CREATE TABLE users_table_local AS SELECT * FROM users_table;
SET client_min_messages TO DEBUG1;
-- CTEs are recursively planned, and subquery foo is also recursively planned
-- final plan becomes a router plan
WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT user_id FROM events_table
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT 
	count(*) 
FROM 
	cte,
	  (SELECT 
    	DISTINCT users_table.user_id 
     FROM 
     	users_table, events_table 
     WHERE 
     	users_table.user_id = events_table.user_id AND 
     event_type IN (1,2,3,4)
     ORDER BY 1 DESC LIMIT 5
     ) as foo 
	  WHERE foo.user_id = cte.user_id;
DEBUG:  generating subplan 2_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 3_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 3_2 for CTE dist_cte: SELECT user_id FROM public.events_table
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 2_2 for subquery SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4]))) ORDER BY users_table.user_id DESC LIMIT 5
 count 
-------
  1644
(1 row)

-- CTEs are recursively planned, and subquery foo is also recursively planned
-- final plan becomes a real-time plan since we also have events_table in the 
-- range table entries 
WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT user_id FROM events_table
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT 
	count(*) 
FROM 
	cte,
	  (SELECT 
    	DISTINCT users_table.user_id 
     FROM 
     	users_table, events_table 
     WHERE 
     	users_table.user_id = events_table.user_id AND 
     event_type IN (1,2,3,4)
     ORDER BY 1 DESC LIMIT 5
     ) as foo, events_table
	  WHERE foo.user_id = cte.user_id AND events_table.user_id = cte.user_id;
DEBUG:  generating subplan 6_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 7_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 7_2 for CTE dist_cte: SELECT user_id FROM public.events_table
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 6_2 for subquery SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4]))) ORDER BY users_table.user_id DESC LIMIT 5
 count 
-------
 30608
(1 row)

-- CTEs are replaced and subquery in WHERE is also replaced
-- but the query is still real-time query since users_table is in the 
-- range table list
WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT user_id FROM events_table
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT DISTINCT cte.user_id
FROM users_table, cte
WHERE 
	users_table.user_id = cte.user_id AND
	users_table.user_id IN (SELECT DISTINCT value_2 FROM users_table WHERE value_1 >= 1 AND value_1 <= 20 ORDER BY 1 LIMIT 5)
    ORDER BY 1 DESC;
DEBUG:  generating subplan 10_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 11_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 11_2 for CTE dist_cte: SELECT user_id FROM public.events_table
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 10_2 for subquery SELECT DISTINCT value_2 FROM public.users_table WHERE ((value_1 >= 1) AND (value_1 <= 20)) ORDER BY value_2 LIMIT 5
 user_id 
---------
       4
       3
       2
       1
(4 rows)

-- a very similar query as the above, but this time errors 
-- out since we don't support subqueries in WHERE clause
-- when there is only intermediate results on the range table
-- note that this time subquery in WHERE clause is not replaced
WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT user_id FROM events_table
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT DISTINCT cte.user_id
FROM cte
WHERE 
	cte.user_id IN (SELECT DISTINCT user_id FROM users_table WHERE value_1 >= 1 AND value_1 <= 20)
    ORDER BY 1 DESC;
DEBUG:  generating subplan 14_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 15_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 15_2 for CTE dist_cte: SELECT user_id FROM public.events_table
ERROR:  cannot pushdown the subquery
DETAIL:  Complex subqueries and CTEs are not allowed in the FROM clause when the query has subqueries in the WHERE clause
-- CTEs inside a subquery and the final query becomes a router
-- query
SELECT
   user_id
FROM
    (
	     WITH cte AS (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	     ) SELECT * FROM cte ORDER BY 1 DESC
     ) as foo;
DEBUG:  generating subplan 17_1 for CTE cte: SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4])))
 user_id 
---------
       6
       5
       4
       3
       2
       1
(6 rows)

-- CTEs inside a subquery and the final query becomes a
-- real-time query since the other subquery is safe to pushdown
SELECT
   bar.user_id
FROM
    (
	     WITH cte AS (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	     ) SELECT * FROM cte ORDER BY 1 DESC
     ) as foo, 
    (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	    
     ) as bar  
WHERE foo.user_id = bar.user_id;
DEBUG:  generating subplan 19_1 for CTE cte: SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4])))
 user_id 
---------
       5
       1
       4
       3
       6
       2
(6 rows)

-- CTEs inside a deeper subquery
-- and also the subquery that contains the CTE is replaced
SELECT
   DISTINCT bar.user_id
FROM
    (
	     WITH cte AS (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	     ) SELECT * FROM cte ORDER BY 1 DESC
     ) as foo, 
    (
	    SELECT 
	    	users_table.user_id, some_events.event_type
	     FROM 
	     	users_table, 
	     	(
	     		WITH cte AS (
			    SELECT 
			    	event_type, users_table.user_id
			     FROM 
			     	users_table, events_table 
			     WHERE 
			     	users_table.user_id = events_table.user_id AND 
			     value_1 IN (1,2)
			     ) SELECT * FROM cte ORDER BY 1 DESC
	     	) as some_events
	     WHERE 
	     	users_table.user_id = some_events.user_id AND 
	     event_type IN (1,2,3,4)
	     ORDER BY 2,1
	     LIMIT 2
	    
     ) as bar  
WHERE foo.user_id = bar.user_id
ORDER BY 1 DESC LIMIT 5;
DEBUG:  generating subplan 21_1 for CTE cte: SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4])))
DEBUG:  generating subplan 21_2 for CTE cte: SELECT events_table.event_type, users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (users_table.value_1 = ANY (ARRAY[1, 2])))
DEBUG:  push down of limit count: 2
DEBUG:  generating subplan 21_3 for subquery SELECT users_table.user_id, some_events.event_type FROM public.users_table, (SELECT cte.event_type, cte.user_id FROM (SELECT intermediate_result.event_type, intermediate_result.user_id FROM read_intermediate_result('21_2'::text, 'binary'::citus_copy_format) intermediate_result(event_type integer, user_id integer)) cte ORDER BY cte.event_type DESC) some_events WHERE ((users_table.user_id = some_events.user_id) AND (some_events.event_type = ANY (ARRAY[1, 2, 3, 4]))) ORDER BY some_events.event_type, users_table.user_id LIMIT 2
 user_id 
---------
       1
(1 row)

-- CTEs on the different parts of the query is replaced 
-- and subquery foo is also replaced since it contains
-- DISTINCT on a non-partition key 
SELECT * FROM 
(
	WITH cte AS (
		WITH local_cte AS (
			SELECT * FROM users_table_local
		),
		dist_cte AS (
			SELECT user_id FROM events_table
		)
		SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
	)
	SELECT DISTINCT cte.user_id
	FROM users_table, cte
	WHERE 
		users_table.user_id = cte.user_id AND
		users_table.user_id IN 
			(WITH cte_in_where AS (SELECT DISTINCT value_2 FROM users_table WHERE value_1 >= 1 AND value_1 <= 20 ORDER BY 1 LIMIT 5) SELECT * FROM cte_in_where)
	    ORDER BY 1 DESC
	    ) as foo, 
			events_table 
		WHERE 
			foo.user_id = events_table.value_2
ORDER BY 3 DESC, 2 DESC, 1 DESC 
LIMIT 5;
DEBUG:  generating subplan 25_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 26_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 26_2 for CTE dist_cte: SELECT user_id FROM public.events_table
DEBUG:  generating subplan 25_2 for CTE cte_in_where: SELECT DISTINCT value_2 FROM public.users_table WHERE ((value_1 >= 1) AND (value_1 <= 20)) ORDER BY value_2 LIMIT 5
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 25_3 for subquery SELECT DISTINCT cte.user_id FROM public.users_table, (SELECT intermediate_result.user_id FROM read_intermediate_result('25_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer)) cte WHERE ((users_table.user_id = cte.user_id) AND (users_table.user_id IN (SELECT cte_in_where.value_2 FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('25_2'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) cte_in_where))) ORDER BY cte.user_id DESC
DEBUG:  push down of limit count: 5
 user_id | user_id |              time               | event_type | value_2 | value_3 | value_4 
---------+---------+---------------------------------+------------+---------+---------+---------
       4 |       1 | Thu Nov 23 21:54:46.924477 2017 |          6 |       4 |       5 |        
       2 |       4 | Thu Nov 23 18:10:21.338399 2017 |          1 |       2 |       4 |        
       4 |       3 | Thu Nov 23 18:08:26.550729 2017 |          2 |       4 |       3 |        
       2 |       3 | Thu Nov 23 16:44:41.903713 2017 |          4 |       2 |       2 |        
       1 |       3 | Thu Nov 23 16:31:56.219594 2017 |          5 |       1 |       2 |        
(5 rows)

-- now recursively plan subqueries inside the CTEs that contains LIMIT and OFFSET
WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT 
			user_id
		FROM 
			events_table, 
			(SELECT DISTINCT value_2 FROM users_table OFFSET 0) as foo
		WHERE 
			events_table.user_id = foo.value_2 AND
			events_table.user_id IN (SELECT DISTINCT value_1 FROM users_table ORDER BY 1 LIMIT 3)
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT 
	count(*) 
FROM 
	cte,
	  (SELECT 
    	DISTINCT users_table.user_id 
     FROM 
     	users_table, events_table 
     WHERE 
     	users_table.user_id = events_table.user_id AND 
     event_type IN (1,2,3,4)
     ORDER BY 1 DESC LIMIT 5
     ) as foo 
	  WHERE foo.user_id = cte.user_id;
DEBUG:  generating subplan 30_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table, (SELECT DISTINCT users_table.value_2 FROM public.users_table OFFSET 0) foo WHERE ((events_table.user_id = foo.value_2) AND (events_table.user_id IN (SELECT DISTINCT users_table.value_1 FROM public.users_table ORDER BY users_table.value_1 LIMIT 3)))) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 31_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 31_2 for CTE dist_cte: SELECT events_table.user_id FROM public.events_table, (SELECT DISTINCT users_table.value_2 FROM public.users_table OFFSET 0) foo WHERE ((events_table.user_id = foo.value_2) AND (events_table.user_id IN (SELECT DISTINCT users_table.value_1 FROM public.users_table ORDER BY users_table.value_1 LIMIT 3)))
DEBUG:  push down of limit count: 3
DEBUG:  generating subplan 32_1 for subquery SELECT DISTINCT value_1 FROM public.users_table ORDER BY value_1 LIMIT 3
DEBUG:  generating subplan 32_2 for subquery SELECT DISTINCT value_2 FROM public.users_table OFFSET 0
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 30_2 for subquery SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4]))) ORDER BY users_table.user_id DESC LIMIT 5
 count 
-------
   432
(1 row)

-- the same query, but this time the CTEs also live inside a subquery
SELECT 
	* 
FROM 
(
	WITH cte AS (
	WITH local_cte AS (
		SELECT * FROM users_table_local
	),
	dist_cte AS (
		SELECT 
			user_id
		FROM 
			events_table, 
			(SELECT DISTINCT value_2 FROM users_table OFFSET 0) as foo
		WHERE 
			events_table.user_id = foo.value_2 AND
			events_table.user_id IN (SELECT DISTINCT value_1 FROM users_table ORDER BY 1 LIMIT 3)
	)
	SELECT dist_cte.user_id FROM local_cte join dist_cte on dist_cte.user_id=local_cte.user_id
)
SELECT 
	count(*)  as cnt
FROM 
	cte,
	  (SELECT 
    	DISTINCT users_table.user_id 
     FROM 
     	users_table, events_table
     WHERE 
     	users_table.user_id = events_table.user_id AND 
     event_type IN (1,2,3,4)
     ORDER BY 1 DESC LIMIT 5
     ) as foo 
	  WHERE foo.user_id = cte.user_id
) as foo, users_table WHERE foo.cnt > users_table.value_2 
ORDER BY 3 DESC, 1 DESC, 2 DESC, 4 DESC
LIMIT 5;
DEBUG:  generating subplan 36_1 for CTE cte: WITH local_cte AS (SELECT users_table_local.user_id, users_table_local."time", users_table_local.value_1, users_table_local.value_2, users_table_local.value_3, users_table_local.value_4 FROM subquery_and_ctes.users_table_local), dist_cte AS (SELECT events_table.user_id FROM public.events_table, (SELECT DISTINCT users_table.value_2 FROM public.users_table OFFSET 0) foo WHERE ((events_table.user_id = foo.value_2) AND (events_table.user_id IN (SELECT DISTINCT users_table.value_1 FROM public.users_table ORDER BY users_table.value_1 LIMIT 3)))) SELECT dist_cte.user_id FROM (local_cte JOIN dist_cte ON ((dist_cte.user_id = local_cte.user_id)))
DEBUG:  generating subplan 37_1 for CTE local_cte: SELECT user_id, "time", value_1, value_2, value_3, value_4 FROM subquery_and_ctes.users_table_local
DEBUG:  generating subplan 37_2 for CTE dist_cte: SELECT events_table.user_id FROM public.events_table, (SELECT DISTINCT users_table.value_2 FROM public.users_table OFFSET 0) foo WHERE ((events_table.user_id = foo.value_2) AND (events_table.user_id IN (SELECT DISTINCT users_table.value_1 FROM public.users_table ORDER BY users_table.value_1 LIMIT 3)))
DEBUG:  push down of limit count: 3
DEBUG:  generating subplan 38_1 for subquery SELECT DISTINCT value_1 FROM public.users_table ORDER BY value_1 LIMIT 3
DEBUG:  generating subplan 38_2 for subquery SELECT DISTINCT value_2 FROM public.users_table OFFSET 0
DEBUG:  push down of limit count: 5
DEBUG:  generating subplan 36_2 for subquery SELECT DISTINCT users_table.user_id FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[1, 2, 3, 4]))) ORDER BY users_table.user_id DESC LIMIT 5
DEBUG:  generating subplan 36_3 for subquery SELECT count(*) AS cnt FROM (SELECT intermediate_result.user_id FROM read_intermediate_result('36_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer)) cte, (SELECT intermediate_result.user_id FROM read_intermediate_result('36_2'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer)) foo WHERE (foo.user_id = cte.user_id)
DEBUG:  push down of limit count: 5
 cnt | user_id |              time               | value_1 | value_2 | value_3 | value_4 
-----+---------+---------------------------------+---------+---------+---------+---------
 432 |       1 | Thu Nov 23 17:30:34.635085 2017 |       3 |       4 |       4 |        
 432 |       1 | Thu Nov 23 17:23:03.441394 2017 |       5 |       4 |       3 |        
 432 |       3 | Thu Nov 23 17:18:51.048758 2017 |       1 |       5 |       5 |        
 432 |       3 | Thu Nov 23 17:10:35.959913 2017 |       4 |       3 |       1 |        
 432 |       5 | Thu Nov 23 16:48:32.08896 2017  |       5 |       2 |       1 |        
(5 rows)

-- recursive CTES are not supported inside subqueries as well
SELECT
   bar.user_id
FROM
    (
	     WITH RECURSIVE cte AS (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	     ) SELECT * FROM cte ORDER BY 1 DESC
     ) as foo, 
    (
	    SELECT 
	    	DISTINCT users_table.user_id 
	     FROM 
	     	users_table, events_table 
	     WHERE 
	     	users_table.user_id = events_table.user_id AND 
	     event_type IN (1,2,3,4)
	    
     ) as bar  
WHERE foo.user_id = bar.user_id;
ERROR:  recursive CTEs are not supported in distributed queries
SET client_min_messages TO DEFAULT;
DROP SCHEMA subquery_and_ctes CASCADE;
NOTICE:  drop cascades to table users_table_local
SET search_path TO public;