--
-- COMPLEX_COUNT_DISTINCT
--
CREATE TABLE lineitem_hash (
	l_orderkey bigint not null,
	l_partkey integer not null,
	l_suppkey integer not null,
	l_linenumber integer not null,
	l_quantity decimal(15, 2) not null,
	l_extendedprice decimal(15, 2) not null,
	l_discount decimal(15, 2) not null,
	l_tax decimal(15, 2) not null,
	l_returnflag char(1) not null,
	l_linestatus char(1) not null,
	l_shipdate date not null,
	l_commitdate date not null,
	l_receiptdate date not null,
	l_shipinstruct char(25) not null,
	l_shipmode char(10) not null,
	l_comment varchar(44) not null,
	PRIMARY KEY(l_orderkey, l_linenumber) );
	
SELECT master_create_distributed_table('lineitem_hash', 'l_orderkey', 'hash');
 master_create_distributed_table 
---------------------------------
 
(1 row)

SELECT master_create_worker_shards('lineitem_hash', 8, 1);
 master_create_worker_shards 
-----------------------------
 
(1 row)

\COPY lineitem_hash FROM '@abs_srcdir@/data/lineitem.1.data' with delimiter '|'
\COPY lineitem_hash FROM '@abs_srcdir@/data/lineitem.2.data' with delimiter '|'
SET citus.task_executor_type to "task-tracker";
-- count(distinct) is supported on top level query if there
-- is a grouping on the partition key	
SELECT
	l_orderkey, count(DISTINCT l_partkey)
	FROM lineitem_hash
	GROUP BY l_orderkey
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_orderkey | count 
------------+-------
      14885 |     7
      14884 |     7
      14821 |     7
      14790 |     7
      14785 |     7
      14755 |     7
      14725 |     7
      14694 |     7
      14627 |     7
      14624 |     7
(10 rows)

-- it is not supported if there is no grouping or grouping is on non-partition field
SELECT
	count(DISTINCT l_partkey)
	FROM lineitem_hash
	ORDER BY 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  table partitioning is unsuitable for aggregate (distinct)
HINT:  You can load the hll extension from contrib packages and enable distinct approximations.
SELECT
	l_shipmode, count(DISTINCT l_partkey)
	FROM lineitem_hash
	GROUP BY l_shipmode
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  table partitioning is unsuitable for aggregate (distinct)
HINT:  You can load the hll extension from contrib packages and enable distinct approximations.
-- count distinct is supported on single table subqueries
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT l_partkey)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_orderkey | count 
------------+-------
      14885 |     7
      14884 |     7
      14821 |     7
      14790 |     7
      14785 |     7
      14755 |     7
      14725 |     7
      14694 |     7
      14627 |     7
      14624 |     7
(10 rows)

SELECT *
	FROM (
		SELECT
			l_partkey, count(DISTINCT l_orderkey)
			FROM lineitem_hash
			GROUP BY l_partkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_partkey | count 
-----------+-------
    199146 |     3
    188804 |     3
    177771 |     3
    160895 |     3
    149926 |     3
    136884 |     3
     87761 |     3
     15283 |     3
      6983 |     3
      1927 |     3
(10 rows)

-- case expr in count distinct is supported.
-- count orders partkeys if l_shipmode is air
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT CASE WHEN l_shipmode = 'AIR' THEN l_partkey ELSE NULL END) as count
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	WHERE count > 0
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_orderkey | count 
------------+-------
      12005 |     4
       5409 |     4
       4964 |     4
      14848 |     3
      14496 |     3
      13473 |     3
      13122 |     3
      12929 |     3
      12645 |     3
      12417 |     3
(10 rows)

-- text like operator is also supported
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT CASE WHEN l_shipmode like '%A%' THEN l_partkey ELSE NULL END) as count
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	WHERE count > 0
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_orderkey | count 
------------+-------
      14275 |     7
      14181 |     7
      13605 |     7
      12707 |     7
      12384 |     7
      11746 |     7
      10727 |     7
      10467 |     7
       5636 |     7
       4614 |     7
(10 rows)

-- count distinct is rejected if it does not reference any columns
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT 1)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  aggregate (distinct) with no columns is unsupported
HINT:  You can load the hll extension from contrib packages and enable distinct approximations.
-- count distinct is rejected if it does not reference any columns
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT (random() * 5)::int)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  aggregate (distinct) with no columns is unsupported
HINT:  You can load the hll extension from contrib packages and enable distinct approximations.
-- even non-const function calls are supported within count distinct
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT (random() * 5)::int = l_linenumber)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 0;
 l_orderkey | count 
------------+-------
(0 rows)

-- multiple nested subquery
SELECT
    total,
    avg(avg_count) as total_avg_count
	FROM (
		SELECT
	        number_sum,
	        count(DISTINCT l_suppkey) as total,
	        avg(total_count) avg_count
    		FROM (
    			SELECT
    				l_suppkey,
					sum(l_linenumber) as number_sum,
					count(DISTINCT l_shipmode) as total_count
					FROM
						lineitem_hash
					WHERE
						l_partkey > 100 and
						l_quantity > 2 and
						l_orderkey < 10000
					GROUP BY
						l_suppkey) as distributed_table
			WHERE
				number_sum >= 10
			GROUP BY
				number_sum) as distributed_table_2
	GROUP BY
		total
	ORDER BY
		total_avg_count DESC;
 total |  total_avg_count   
-------+--------------------
     1 | 3.6000000000000000
     6 | 2.8333333333333333
    10 | 2.6000000000000000
    27 | 2.5555555555555556
    32 | 2.4687500000000000
    77 | 2.1948051948051948
    57 | 2.1754385964912281
(7 rows)

-- multiple cases query
SELECT *
	FROM (
		SELECT
			count(DISTINCT
				CASE
					WHEN l_shipmode = 'TRUCK' THEN l_partkey
					WHEN l_shipmode = 'AIR' THEN l_quantity
					WHEN l_shipmode = 'SHIP' THEN l_discount
					ELSE l_suppkey
				END) as count,
			l_shipdate
		FROM
			lineitem_hash
		GROUP BY
			l_shipdate) sub
	WHERE
		count > 0
	ORDER BY
		1 DESC, 2 DESC
	LIMIT 10;
 count | l_shipdate 
-------+------------
    14 | 07-30-1997
    13 | 05-26-1998
    13 | 08-08-1997
    13 | 11-17-1995
    13 | 01-09-1993
    12 | 01-15-1998
    12 | 10-15-1997
    12 | 09-07-1997
    12 | 06-02-1997
    12 | 03-14-1997
(10 rows)

-- count DISTINCT expression
SELECT *
	FROM (
		SELECT
			l_quantity, count(DISTINCT ((l_orderkey / 1000)  * 1000 ))  as count
			FROM
				lineitem_hash
			GROUP BY
				l_quantity) sub
	WHERE
		count > 0
	ORDER BY
		2 DESC, 1 DESC
	LIMIT 10;
 l_quantity | count 
------------+-------
      48.00 |    13
      47.00 |    13
      37.00 |    13
      33.00 |    13
      26.00 |    13
      25.00 |    13
      23.00 |    13
      21.00 |    13
      15.00 |    13
      12.00 |    13
(10 rows)

-- count DISTINCT is part of an expression which inclues another aggregate
SELECT *
	FROM (
		SELECT
			sum(((l_partkey * l_tax) / 100)) /
				count(DISTINCT
					CASE
						WHEN l_shipmode = 'TRUCK' THEN l_partkey
						ELSE l_suppkey
					END) as avg,
			l_shipmode
			FROM
				lineitem_hash
			GROUP BY
				l_shipmode) sub
	ORDER BY
		1 DESC, 2 DESC
	LIMIT 10;
           avg           | l_shipmode 
-------------------------+------------
 44.82904609027336300064 | MAIL      
 44.80704536679536679537 | SHIP      
 44.68891732736572890026 | AIR       
 44.34106724470134874759 | REG AIR   
 43.12739987269255251432 | FOB       
 43.07299253636938646426 | RAIL      
 40.50298377916903813318 | TRUCK     
(7 rows)

--- count DISTINCT CASE WHEN expression
SELECT *
	FROM (
		SELECT
			count(DISTINCT
				CASE
					WHEN l_shipmode = 'TRUCK' THEN l_linenumber
					WHEN l_shipmode = 'AIR' THEN l_linenumber + 10
					ELSE 2
				END) as avg
			FROM
				lineitem_hash
			GROUP BY  l_shipdate) sub
	ORDER BY 1 DESC
	LIMIT 10;
 avg 
-----
   7
   6
   6
   6
   6
   6
   6
   6
   5
   5
(10 rows)

-- COUNT DISTINCT (c1, c2)
SELECT *
	FROM
		(SELECT
			l_shipmode,
			count(DISTINCT (l_shipdate, l_tax))
			FROM
				lineitem_hash
			GROUP BY
				l_shipmode) t
	ORDER BY
		2 DESC,1 DESC
	LIMIT 10;
 l_shipmode | count 
------------+-------
 TRUCK      |  1689
 MAIL       |  1683
 FOB        |  1655
 AIR        |  1650
 SHIP       |  1644
 RAIL       |  1636
 REG AIR    |  1607
(7 rows)

-- other distinct aggregate are not supported
SELECT *
	FROM (
		SELECT
			l_orderkey, sum(DISTINCT l_partkey)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  Only count(distinct) aggregate is supported in subqueries
SELECT *
	FROM (
		SELECT
			l_orderkey, avg(DISTINCT l_partkey)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute aggregate (distinct)
DETAIL:  Only count(distinct) aggregate is supported in subqueries
-- whole row references, oid, and ctid are not supported in count distinct
-- test table does not have oid or ctid enabled, so tests for them are skipped
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT lineitem_hash)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute count (distinct)
DETAIL:  Non-column references are not supported yet
SELECT *
	FROM (
		SELECT
			l_orderkey, count(DISTINCT lineitem_hash.*)
			FROM lineitem_hash
			GROUP BY l_orderkey) sub
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
ERROR:  cannot compute count (distinct)
DETAIL:  Non-column references are not supported yet
DROP TABLE lineitem_hash;