Add some more basic regression tests, mostly for documentation purposes

src/test/regress/expected/subquery_basics.out

@@ -83,6 +83,73 @@ DEBUG:  Plan 5 query after replacing subqueries and CTEs: SELECT value_2, avg FR
        0 | 2.0940170940170940
 (6 rows)
 
+-- subqueries with only generate_series
+SELECT
+   *
+FROM
+    (SELECT 
+        events_table.value_2
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+     ORDER BY 1 DESC
+     LIMIT 5
+     ) as foo,
+    (
+        SELECT i FROM generate_series(0, 100) i
+    ) as bar
+    WHERE foo.value_2 = bar.i
+    ORDER BY 2 DESC, 1;
+DEBUG:  push down of limit count: 5
+DEBUG:  generating subplan 7_1 for subquery SELECT value_2 FROM public.events_table WHERE (event_type = ANY (ARRAY[1, 2, 3, 4])) ORDER BY value_2 DESC LIMIT 5
+DEBUG:  Plan 7 query after replacing subqueries and CTEs: SELECT foo.value_2, bar.i FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('7_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo, (SELECT i.i FROM generate_series(0, 100) i(i)) bar WHERE (foo.value_2 = bar.i) ORDER BY bar.i DESC, foo.value_2
+ value_2 | i 
+---------+---
+       5 | 5
+       5 | 5
+       5 | 5
+       5 | 5
+       5 | 5
+(5 rows)
+
+-- subquery with aggregates without GROUP BY
+SELECT
+   *
+FROM
+    (SELECT 
+        count(*)
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+     ) as foo;
+DEBUG:  generating subplan 9_1 for subquery SELECT count(*) AS count FROM public.events_table WHERE (event_type = ANY (ARRAY[1, 2, 3, 4]))
+DEBUG:  Plan 9 query after replacing subqueries and CTEs: SELECT count FROM (SELECT intermediate_result.count FROM read_intermediate_result('9_1'::text, 'binary'::citus_copy_format) intermediate_result(count bigint)) foo
+ count 
+-------
+    87
+(1 row)
+
+-- subquery having without GROUP BY
+SELECT
+   *
+FROM
+    (SELECT 
+          SUM(events_table.user_id) 
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+    HAVING 
+        MIN(value_2) > 2
+     ) as foo;
+DEBUG:  generating subplan 11_1 for subquery SELECT sum(user_id) AS sum FROM public.events_table WHERE (event_type = ANY (ARRAY[1, 2, 3, 4])) HAVING (min(value_2) > 2)
+DEBUG:  Plan 11 query after replacing subqueries and CTEs: SELECT sum FROM (SELECT intermediate_result.sum FROM read_intermediate_result('11_1'::text, 'binary'::citus_copy_format) intermediate_result(sum bigint)) foo
+ sum 
+-----
+(0 rows)
+
 -- multiple subqueries in FROM clause should be replaced
 -- and the final query is router query
 SELECT
@@ -110,9 +177,9 @@ FROM
      ) as bar
 	WHERE foo.value_2 = bar.value_3
 	ORDER BY 2 DESC, 1;
-DEBUG:  generating subplan 7_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
+DEBUG:  generating subplan 13_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
-DEBUG:  generating subplan 7_2 for subquery SELECT users_table.value_3 FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[5, 6, 7, 8]))) GROUP BY users_table.value_3 ORDER BY users_table.value_3 DESC
+DEBUG:  generating subplan 13_2 for subquery SELECT users_table.value_3 FROM public.users_table, public.events_table WHERE ((users_table.user_id = events_table.user_id) AND (events_table.event_type = ANY (ARRAY[5, 6, 7, 8]))) GROUP BY users_table.value_3 ORDER BY users_table.value_3 DESC
-DEBUG:  Plan 7 query after replacing subqueries and CTEs: SELECT foo.value_2, bar.value_3 FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('7_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo, (SELECT intermediate_result.value_3 FROM read_intermediate_result('7_2'::text, 'binary'::citus_copy_format) intermediate_result(value_3 double precision)) bar WHERE ((foo.value_2)::double precision = bar.value_3) ORDER BY bar.value_3 DESC, foo.value_2
+DEBUG:  Plan 13 query after replacing subqueries and CTEs: SELECT foo.value_2, bar.value_3 FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('13_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo, (SELECT intermediate_result.value_3 FROM read_intermediate_result('13_2'::text, 'binary'::citus_copy_format) intermediate_result(value_3 double precision)) bar WHERE ((foo.value_2)::double precision = bar.value_3) ORDER BY bar.value_3 DESC, foo.value_2
  value_2 | value_3 
 ---------+---------
        5 |       5
@@ -149,8 +216,8 @@ FROM
     WHERE foo.postgres = bar.citus
     ORDER BY 1 DESC, 2 DESC
     LIMIT 3;
-DEBUG:  generating subplan 10_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
+DEBUG:  generating subplan 16_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
-DEBUG:  Plan 10 query after replacing subqueries and CTEs: SELECT DISTINCT ON (bar.citus) bar.citus, foo.postgres, (bar.citus + 1) AS c1, (foo.postgres - 1) AS p1 FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('10_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo(postgres), (SELECT 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[5, 6, 7, 8]))) ORDER BY users_table.user_id DESC) bar(citus) WHERE (foo.postgres = bar.citus) ORDER BY bar.citus DESC, foo.postgres DESC LIMIT 3
+DEBUG:  Plan 16 query after replacing subqueries and CTEs: SELECT DISTINCT ON (bar.citus) bar.citus, foo.postgres, (bar.citus + 1) AS c1, (foo.postgres - 1) AS p1 FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('16_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo(postgres), (SELECT 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[5, 6, 7, 8]))) ORDER BY users_table.user_id DESC) bar(citus) WHERE (foo.postgres = bar.citus) ORDER BY bar.citus DESC, foo.postgres DESC LIMIT 3
 DEBUG:  push down of limit count: 3
  citus | postgres | c1 | p1 
 -------+----------+----+----
@@ -186,8 +253,8 @@ FROM
 	WHERE foo.value_2 = bar.user_id
 	ORDER BY 2 DESC, 1 DESC
 	LIMIT 3;
-DEBUG:  generating subplan 12_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
+DEBUG:  generating subplan 18_1 for subquery SELECT users_table.value_2 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]))) GROUP BY users_table.value_2 ORDER BY users_table.value_2 DESC
-DEBUG:  Plan 12 query after replacing subqueries and CTEs: SELECT foo.value_2, bar.user_id FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('12_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo, (SELECT 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[5, 6, 7, 8]))) ORDER BY users_table.user_id DESC) bar WHERE (foo.value_2 = bar.user_id) ORDER BY bar.user_id DESC, foo.value_2 DESC LIMIT 3
+DEBUG:  Plan 18 query after replacing subqueries and CTEs: SELECT foo.value_2, bar.user_id FROM (SELECT intermediate_result.value_2 FROM read_intermediate_result('18_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer)) foo, (SELECT 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[5, 6, 7, 8]))) ORDER BY users_table.user_id DESC) bar WHERE (foo.value_2 = bar.user_id) ORDER BY bar.user_id DESC, foo.value_2 DESC LIMIT 3
 DEBUG:  push down of limit count: 3
  value_2 | user_id 
 ---------+---------
@@ -203,8 +270,8 @@ WHERE
 	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:  push down of limit count: 5
-DEBUG:  generating subplan 14_1 for subquery SELECT DISTINCT value_2 FROM public.users_table WHERE ((value_1 >= 1) AND (value_1 <= 20)) ORDER BY value_2 LIMIT 5
+DEBUG:  generating subplan 20_1 for subquery SELECT DISTINCT value_2 FROM public.users_table WHERE ((value_1 >= 1) AND (value_1 <= 20)) ORDER BY value_2 LIMIT 5
-DEBUG:  Plan 14 query after replacing subqueries and CTEs: SELECT DISTINCT user_id FROM public.users_table WHERE (user_id IN (SELECT intermediate_result.value_2 FROM read_intermediate_result('14_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer))) ORDER BY user_id DESC
+DEBUG:  Plan 20 query after replacing subqueries and CTEs: SELECT DISTINCT user_id FROM public.users_table WHERE (user_id IN (SELECT intermediate_result.value_2 FROM read_intermediate_result('20_1'::text, 'binary'::citus_copy_format) intermediate_result(value_2 integer))) ORDER BY user_id DESC
  user_id 
 ---------
        4
@@ -232,8 +299,8 @@ FROM
 	) as sub1
 	ORDER BY 1 DESC 
 	LIMIT 3;
-DEBUG:  generating subplan 16_1 for subquery SELECT user_id, event_type FROM public.events_table WHERE (value_2 < 3) OFFSET 3
+DEBUG:  generating subplan 22_1 for subquery SELECT user_id, event_type FROM public.events_table WHERE (value_2 < 3) OFFSET 3
-DEBUG:  Plan 16 query after replacing subqueries and CTEs: SELECT DISTINCT user_id FROM (SELECT users_table.user_id FROM public.users_table, (SELECT bar.event_type, bar.user_id FROM (SELECT foo.event_type, users_table_1.user_id FROM public.users_table users_table_1, (SELECT intermediate_result.user_id, intermediate_result.event_type FROM read_intermediate_result('16_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer, event_type integer)) foo WHERE (foo.user_id = users_table_1.user_id)) bar) baz WHERE (baz.user_id = users_table.user_id)) sub1 ORDER BY user_id DESC LIMIT 3
+DEBUG:  Plan 22 query after replacing subqueries and CTEs: SELECT DISTINCT user_id FROM (SELECT users_table.user_id FROM public.users_table, (SELECT bar.event_type, bar.user_id FROM (SELECT foo.event_type, users_table_1.user_id FROM public.users_table users_table_1, (SELECT intermediate_result.user_id, intermediate_result.event_type FROM read_intermediate_result('22_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer, event_type integer)) foo WHERE (foo.user_id = users_table_1.user_id)) bar) baz WHERE (baz.user_id = users_table.user_id)) sub1 ORDER BY user_id DESC LIMIT 3
 DEBUG:  push down of limit count: 3
  user_id 
 ---------
@@ -269,8 +336,8 @@ FROM (
 ) q
 ORDER BY 2 DESC, 1;
 DEBUG:  push down of limit count: 5
-DEBUG:  generating subplan 18_1 for subquery SELECT user_id FROM public.users_table WHERE ((value_2 >= 5) AND (EXISTS (SELECT events_table.user_id FROM public.events_table WHERE ((events_table.event_type > 1) AND (events_table.event_type <= 3) AND (events_table.value_3 > (1)::double precision) AND (events_table.user_id = users_table.user_id)))) AND (NOT (EXISTS (SELECT events_table.user_id FROM public.events_table WHERE ((events_table.event_type > 3) AND (events_table.event_type <= 4) AND (events_table.value_3 > (1)::double precision) AND (events_table.user_id = users_table.user_id)))))) LIMIT 5
+DEBUG:  generating subplan 24_1 for subquery SELECT user_id FROM public.users_table WHERE ((value_2 >= 5) AND (EXISTS (SELECT events_table.user_id FROM public.events_table WHERE ((events_table.event_type > 1) AND (events_table.event_type <= 3) AND (events_table.value_3 > (1)::double precision) AND (events_table.user_id = users_table.user_id)))) AND (NOT (EXISTS (SELECT events_table.user_id FROM public.events_table WHERE ((events_table.event_type > 3) AND (events_table.event_type <= 4) AND (events_table.value_3 > (1)::double precision) AND (events_table.user_id = users_table.user_id)))))) LIMIT 5
-DEBUG:  Plan 18 query after replacing subqueries and CTEs: SELECT user_id, array_length(events_table, 1) AS array_length FROM (SELECT t.user_id, array_agg(t.event ORDER BY t."time") AS events_table FROM (SELECT u.user_id, (e.event_type)::text AS event, e."time" FROM public.users_table u, public.events_table e WHERE ((u.user_id = e.user_id) AND (u.user_id IN (SELECT intermediate_result.user_id FROM read_intermediate_result('18_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer))))) t GROUP BY t.user_id) q ORDER BY (array_length(events_table, 1)) DESC, user_id
+DEBUG:  Plan 24 query after replacing subqueries and CTEs: SELECT user_id, array_length(events_table, 1) AS array_length FROM (SELECT t.user_id, array_agg(t.event ORDER BY t."time") AS events_table FROM (SELECT u.user_id, (e.event_type)::text AS event, e."time" FROM public.users_table u, public.events_table e WHERE ((u.user_id = e.user_id) AND (u.user_id IN (SELECT intermediate_result.user_id FROM read_intermediate_result('24_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer))))) t GROUP BY t.user_id) q ORDER BY (array_length(events_table, 1)) DESC, user_id
  user_id | array_length 
 ---------+--------------
        5 |          364
@@ -347,8 +414,8 @@ GROUP BY user_id
 HAVING count(*) > 1 AND sum(value_2) > 29
 ORDER BY 1;
 DEBUG:  push down of limit count: 10
-DEBUG:  generating subplan 20_1 for subquery SELECT user_id, count(*) AS count_pay FROM public.users_table WHERE ((user_id >= 1) AND (user_id <= 3) AND (value_1 > 3) AND (value_1 < 5)) GROUP BY user_id HAVING (count(*) > 1) LIMIT 10
+DEBUG:  generating subplan 26_1 for subquery SELECT user_id, count(*) AS count_pay FROM public.users_table WHERE ((user_id >= 1) AND (user_id <= 3) AND (value_1 > 3) AND (value_1 < 5)) GROUP BY user_id HAVING (count(*) > 1) LIMIT 10
-DEBUG:  Plan 20 query after replacing subqueries and CTEs: SELECT user_id FROM public.users_table WHERE (user_id IN (SELECT subquery_top.user_id FROM (SELECT subquery_1.user_id, subquery_2.count_pay FROM ((SELECT users_table_1.user_id, 'action=>1'::text AS event, events_table."time" FROM public.users_table users_table_1, public.events_table WHERE ((users_table_1.user_id = events_table.user_id) AND (users_table_1.user_id >= 1) AND (users_table_1.user_id <= 3) AND (events_table.event_type > 1) AND (events_table.event_type < 3)) UNION SELECT users_table_1.user_id, 'action=>2'::text AS event, events_table."time" FROM public.users_table users_table_1, public.events_table WHERE ((users_table_1.user_id = events_table.user_id) AND (users_table_1.user_id >= 1) AND (users_table_1.user_id <= 3) AND (events_table.event_type > 2) AND (events_table.event_type < 4))) subquery_1 LEFT JOIN (SELECT intermediate_result.user_id, intermediate_result.count_pay FROM read_intermediate_result('20_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer, count_pay bigint)) subquery_2 ON ((subquery_1.user_id = subquery_2.user_id))) GROUP BY subquery_1.user_id, subquery_2.count_pay) subquery_top GROUP BY subquery_top.count_pay, subquery_top.user_id)) GROUP BY user_id HAVING ((count(*) > 1) AND (sum(value_2) > 29)) ORDER BY user_id
+DEBUG:  Plan 26 query after replacing subqueries and CTEs: SELECT user_id FROM public.users_table WHERE (user_id IN (SELECT subquery_top.user_id FROM (SELECT subquery_1.user_id, subquery_2.count_pay FROM ((SELECT users_table_1.user_id, 'action=>1'::text AS event, events_table."time" FROM public.users_table users_table_1, public.events_table WHERE ((users_table_1.user_id = events_table.user_id) AND (users_table_1.user_id >= 1) AND (users_table_1.user_id <= 3) AND (events_table.event_type > 1) AND (events_table.event_type < 3)) UNION SELECT users_table_1.user_id, 'action=>2'::text AS event, events_table."time" FROM public.users_table users_table_1, public.events_table WHERE ((users_table_1.user_id = events_table.user_id) AND (users_table_1.user_id >= 1) AND (users_table_1.user_id <= 3) AND (events_table.event_type > 2) AND (events_table.event_type < 4))) subquery_1 LEFT JOIN (SELECT intermediate_result.user_id, intermediate_result.count_pay FROM read_intermediate_result('26_1'::text, 'binary'::citus_copy_format) intermediate_result(user_id integer, count_pay bigint)) subquery_2 ON ((subquery_1.user_id = subquery_2.user_id))) GROUP BY subquery_1.user_id, subquery_2.count_pay) subquery_top GROUP BY subquery_top.count_pay, subquery_top.user_id)) GROUP BY user_id HAVING ((count(*) > 1) AND (sum(value_2) > 29)) ORDER BY user_id
  user_id 
 ---------
        2

src/test/regress/sql/subquery_basics.sql

@@ -53,6 +53,51 @@ FROM
      ) as foo
     ORDER BY 2 DESC, 1;
 
+-- subqueries with only generate_series
+SELECT
+   *
+FROM
+    (SELECT 
+        events_table.value_2
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+     ORDER BY 1 DESC
+     LIMIT 5
+     ) as foo,
+    (
+        SELECT i FROM generate_series(0, 100) i
+    ) as bar
+    WHERE foo.value_2 = bar.i
+    ORDER BY 2 DESC, 1;
+
+-- subquery with aggregates without GROUP BY
+SELECT
+   *
+FROM
+    (SELECT 
+        count(*)
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+     ) as foo;
+
+-- subquery having without GROUP BY
+SELECT
+   *
+FROM
+    (SELECT 
+          SUM(events_table.user_id) 
+     FROM 
+        events_table 
+    WHERE
+     event_type IN (1,2,3,4)
+    HAVING 
+        MIN(value_2) > 2
+     ) as foo;
+
 -- multiple subqueries in FROM clause should be replaced
 -- and the final query is router query
 SELECT