--
-- MULTI_AGG_APPROXIMATE_DISTINCT
--
-- Create HLL extension if present, print false result otherwise
SELECT CASE WHEN COUNT(*) > 0 THEN
	'CREATE EXTENSION HLL'
ELSE 'SELECT false AS hll_present' END
AS create_cmd FROM pg_available_extensions()
WHERE name = 'hll'
\gset
:create_cmd;
-- Try to execute count(distinct) when approximate distincts aren't enabled
SELECT count(distinct l_orderkey) FROM lineitem;
 count 
-------
  2985
(1 row)

-- Check approximate count(distinct) at different precisions / error rates
SET citus.count_distinct_error_rate = 0.1;
SELECT count(distinct l_orderkey) FROM lineitem;
 count 
-------
  2612
(1 row)

SET citus.count_distinct_error_rate = 0.01;
SELECT count(distinct l_orderkey) FROM lineitem;
 count 
-------
  2967
(1 row)

-- Check approximate count(distinct) for different data types
SELECT count(distinct l_partkey) FROM lineitem;
 count 
-------
 11654
(1 row)

SELECT count(distinct l_extendedprice) FROM lineitem;
 count 
-------
 11691
(1 row)

SELECT count(distinct l_shipdate) FROM lineitem;
 count 
-------
  2483
(1 row)

SELECT count(distinct l_comment) FROM lineitem;
 count 
-------
 11788
(1 row)

-- Check that we can execute approximate count(distinct) on complex expressions
SELECT count(distinct (l_orderkey * 2 + 1)) FROM lineitem;
 count 
-------
  2980
(1 row)

SELECT count(distinct extract(month from l_shipdate)) AS my_month FROM lineitem;
 my_month 
----------
       12
(1 row)

SELECT count(distinct l_partkey) / count(distinct l_orderkey) FROM lineitem;
 ?column? 
----------
        3
(1 row)

-- Check that we can execute approximate count(distinct) on select queries that
-- contain different filter, join, sort and limit clauses
SELECT count(distinct l_orderkey) FROM lineitem
	WHERE octet_length(l_comment) + octet_length('randomtext'::text) > 40;
 count 
-------
  2355
(1 row)

SELECT count(DISTINCT l_orderkey) FROM lineitem, orders
	WHERE l_orderkey = o_orderkey AND l_quantity < 5;
 count 
-------
   835
(1 row)

SELECT count(DISTINCT l_orderkey) as distinct_order_count, l_quantity FROM lineitem
	WHERE l_quantity < 32.0
	GROUP BY l_quantity
	ORDER BY distinct_order_count ASC, l_quantity ASC
	LIMIT 10;
 distinct_order_count | l_quantity 
----------------------+------------
                  210 |      29.00
                  216 |      13.00
                  217 |      16.00
                  219 |       3.00
                  220 |      18.00
                  222 |      14.00
                  223 |       7.00
                  223 |      17.00
                  223 |      26.00
                  223 |      31.00
(10 rows)

-- Check that approximate count(distinct) works at a table in a schema other than public
-- create necessary objects
SET citus.next_shard_id TO 20000000;
SET citus.next_placement_id TO 20000000;
CREATE SCHEMA test_count_distinct_schema;
CREATE TABLE test_count_distinct_schema.nation_hash(
    n_nationkey integer not null,
    n_name char(25) not null,
    n_regionkey integer not null,
    n_comment varchar(152)
);
SELECT create_distributed_table('test_count_distinct_schema.nation_hash', 'n_nationkey', 'hash');
 create_distributed_table 
--------------------------
 
(1 row)

\copy test_count_distinct_schema.nation_hash FROM STDIN with delimiter '|';
SET search_path TO public;
SET citus.count_distinct_error_rate TO 0.01;
SELECT COUNT (DISTINCT n_regionkey) FROM test_count_distinct_schema.nation_hash;
 count 
-------
     3
(1 row)

-- test with search_path is set
SET search_path TO test_count_distinct_schema;
SELECT COUNT (DISTINCT n_regionkey) FROM nation_hash;
 count 
-------
     3
(1 row)

SET search_path TO public;
-- If we have an order by on count(distinct) that we intend to push down to
-- worker nodes, we need to error out. Otherwise, we are fine.
SET citus.limit_clause_row_fetch_count = 1000;
SELECT l_returnflag, count(DISTINCT l_shipdate) as count_distinct, count(*) as total
	FROM lineitem
	GROUP BY l_returnflag
	ORDER BY count_distinct
	LIMIT 10;
ERROR:  cannot approximate count(distinct) and order by it
HINT:  You might need to disable approximations for either count(distinct) or limit through configuration.
SELECT l_returnflag, count(DISTINCT l_shipdate) as count_distinct, count(*) as total
	FROM lineitem
	GROUP BY l_returnflag
	ORDER BY total
	LIMIT 10;
 l_returnflag | count_distinct | total 
--------------+----------------+-------
 R            |           1103 |  2901
 A            |           1108 |  2944
 N            |           1265 |  6155
(3 rows)

SELECT
	l_orderkey,
	count(l_partkey) FILTER (WHERE l_shipmode = 'AIR'),
	count(DISTINCT l_partkey) FILTER (WHERE l_shipmode = 'AIR'),
	count(DISTINCT CASE WHEN l_shipmode = 'AIR' THEN l_partkey ELSE NULL END)
	FROM lineitem
	GROUP BY l_orderkey
	ORDER BY 2 DESC, 1 DESC
	LIMIT 10;
 l_orderkey | count | count | count 
------------+-------+-------+-------
      12005 |     4 |     4 |     4
       5409 |     4 |     4 |     4
       4964 |     4 |     4 |     4
      14848 |     3 |     3 |     3
      14496 |     3 |     3 |     3
      13473 |     3 |     3 |     3
      13122 |     3 |     3 |     3
      12929 |     3 |     3 |     3
      12645 |     3 |     3 |     3
      12417 |     3 |     3 |     3
(10 rows)

-- Check that we can revert config and disable count(distinct) approximations
SET citus.count_distinct_error_rate = 0.0;
SELECT count(distinct l_orderkey) FROM lineitem;
 count 
-------
  2985
(1 row)