--
-- MULTI_SINGLE_RELATION_SUBQUERY
--
-- This test checks that we are able to run selected set of distributed SQL subqueries.
SET citus.next_shard_id TO 860000;
SET citus.task_executor_type TO 'task-tracker';
select
    number_sum,
    count(*) as total,
    avg(total_count) avg_count
from
    (select
        l_suppkey,
        l_linestatus,
        sum(l_linenumber) as number_sum,
        count(*) as total_count
    from
        lineitem
    group by
        l_suppkey,
        l_linestatus) as distributed_table
where
    number_sum >= 10
group by
    number_sum
order by
    total desc,
    number_sum desc
limit 10;
 number_sum | total |     avg_count      
------------+-------+--------------------
         10 |   136 | 2.3970588235294118
         11 |    97 | 2.6082474226804124
         12 |    56 | 2.8392857142857143
         13 |    42 | 2.8809523809523810
         14 |    21 | 3.2857142857142857
         16 |    10 | 3.5000000000000000
         15 |    10 | 3.3000000000000000
         17 |     6 | 3.3333333333333333
         18 |     3 | 4.0000000000000000
         19 |     2 | 4.0000000000000000
(10 rows)

-- same query above, just replace outer where clause with inner having clause
select
    number_sum,
    count(*) as total,
    avg(total_count) avg_count
from
    (select
        l_suppkey,
        l_linestatus,
        sum(l_linenumber) as number_sum,
        count(*) as total_count
    from
        lineitem
    group by
        l_suppkey,
        l_linestatus
    having
        sum(l_linenumber) >= 10) as distributed_table
group by
    number_sum
order by
    total desc,
    number_sum desc
limit 10;
 number_sum | total |     avg_count      
------------+-------+--------------------
         10 |   136 | 2.3970588235294118
         11 |    97 | 2.6082474226804124
         12 |    56 | 2.8392857142857143
         13 |    42 | 2.8809523809523810
         14 |    21 | 3.2857142857142857
         16 |    10 | 3.5000000000000000
         15 |    10 | 3.3000000000000000
         17 |     6 | 3.3333333333333333
         18 |     3 | 4.0000000000000000
         19 |     2 | 4.0000000000000000
(10 rows)

select
    (l_suppkey / 100) as suppkey_bin,
    avg(total_count) avg_count
from
    (select
        l_suppkey,
        sum(l_linenumber) as number_sum,
        count(*) as total_count
    from
        lineitem
    group by
        l_suppkey,
        l_linestatus) as distributed_table
group by
    suppkey_bin
order by
    avg_count desc, suppkey_bin DESC
limit 20;
 suppkey_bin |     avg_count      
-------------+--------------------
          95 | 1.4851485148514851
          90 | 1.4761904761904762
          52 | 1.4680851063829787
          40 | 1.4659090909090909
          15 | 1.4642857142857143
          75 | 1.4444444444444444
          72 | 1.4375000000000000
          84 | 1.4242424242424242
          35 | 1.4226804123711340
          64 | 1.4166666666666667
          74 | 1.4117647058823529
          21 | 1.4000000000000000
          18 | 1.4000000000000000
          26 | 1.3932584269662921
          96 | 1.3913043478260870
          71 | 1.3913043478260870
          86 | 1.3894736842105263
          55 | 1.3882352941176471
          57 | 1.3875000000000000
           1 | 1.3846153846153846
(20 rows)

select
    total,
    avg(avg_count) as total_avg_count
from
    (select
        number_sum,
        count(*) as total,
        avg(total_count) avg_count
    from
        (select
            l_suppkey,
            sum(l_linenumber) as number_sum,
            count(*) as total_count
        from
            lineitem
        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;
 total |  total_avg_count   
-------+--------------------
     1 | 4.8000000000000000
     6 | 3.0000000000000000
    10 | 3.5000000000000000
    27 | 2.9259259259259259
    32 | 2.8125000000000000
    57 | 2.4912280701754386
    77 | 2.3896103896103896
(7 rows)

-- Check that we support subquery even though group by clause is an expression
-- and it is not referred in the target list.
select
    avg(count)
from
    (select
        l_suppkey,
        count(*) as count
    from
        lineitem
    group by
        (l_orderkey/4)::int,
        l_suppkey )  as distributed_table;
          avg           
------------------------
 1.00083402835696413678
(1 row)

-- we don't support subqueries with limit.
select
    l_suppkey,
    sum(suppkey_count) as total_suppkey_count
from
    (select
        l_suppkey,
        count(*) as suppkey_count
    from
        lineitem
    group by
        l_suppkey
    order by
        l_suppkey
    limit 100) as distributed_table
group by
    l_suppkey
    ORDER BY 2 DESC, 1 DESC 
LIMIT 5;
ERROR:  cannot perform distributed planning on this query
DETAIL:  Subqueries with limit are not supported yet
-- Check that we don't support subqueries without aggregates.
select
    DISTINCT rounded_tax
from
    (select
        round(l_tax) as rounded_tax
    from
        lineitem
    group by
        l_tax) as distributed_table
    ORDER BY 1 DESC
    LIMIT 5;
ERROR:  cannot perform distributed planning on this query
DETAIL:  Subqueries without aggregates are not supported yet
-- Check that we support subqueries with count(distinct).
select
    avg(different_shipment_days)
from
    (select
        count(distinct l_shipdate) as different_shipment_days
    from
        lineitem
    group by
        l_partkey) as distributed_table;
          avg           
------------------------
 1.02907126318497555956
(1 row)

select
    avg(different_shipment_days)
from
    (select
        count(distinct l_shipdate) as different_shipment_days
    from
        lineitem
    group by
        l_partkey
    having 
        count(distinct l_shipdate) >= 2) as distributed_table;
        avg         
--------------------
 2.0335365853658537
(1 row)

-- Check that if subquery is pulled, we don't error and run query properly.
SELECT max(l_suppkey) FROM
(
    SELECT
        l_suppkey
        FROM (
            SELECT
                l_suppkey,
                count(*)
            FROM
                lineitem
            WHERE
                l_orderkey < 20000
            GROUP BY
                l_suppkey) z
) y;
 max  
------
 9999
(1 row)