external
/
citus
mirror of https://github.com/citusdata/citus.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #1668 from citusdata/window_function_preliminary_implementation 

Add window function support for SUBQUERY PUSHDOWN and INSERT INTO SELECT
pull/1699/head
Mehmet Furkan ŞAHİN 2017-10-04 17:18:26 +03:00 committed by GitHub
parent f6b43d81ec 498ac80d8b
commit e202c51fec
15 changed files with 4091 additions and 627 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
src/backend/distributed/planner/insert_select_planner.c
View File

@ -776,6 +776,7 @@ MultiTaskRouterSelectQuerySupported(Query *query)
{
List *queryList = NIL;
ListCell *queryCell = NULL;
StringInfo errorDetail = NULL;
ExtractQueryWalker((Node *) query, &queryList);
foreach(queryCell, queryList)
@ -797,7 +798,7 @@ MultiTaskRouterSelectQuerySupported(Query *query)
if (subquery->limitCount != NULL)
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"LIMIT clauses are not allowed in distirbuted INSERT "
"LIMIT clauses are not allowed in distributed INSERT "
"... SELECT queries",
NULL, NULL);
}
@ -811,18 +812,35 @@ MultiTaskRouterSelectQuerySupported(Query *query)
NULL, NULL);
}
/*
* We could potentially support window clauses where the data is partitioned
* over distribution column. For simplicity, we currently do not support window
* clauses at all.
*/
if (subquery->windowClause != NULL)
/* group clause list must include partition column */
if (subquery->groupClause)
{
List *groupClauseList = subquery->groupClause;
List *targetEntryList = subquery->targetList;
List *groupTargetEntryList = GroupTargetEntryList(groupClauseList,
targetEntryList);
bool groupOnPartitionColumn = TargetListOnPartitionColumn(subquery,
groupTargetEntryList);
if (!groupOnPartitionColumn)
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED,
"window functions are not allowed in distributed "
"INSERT ... SELECT queries",
"Group by list without distribution column is "
"not allowed in distributed INSERT ... "
"SELECT queries",
NULL, NULL);
}
}
/*
* We support window functions when the window function
* is partitioned on distribution column.
*/
if (subquery->windowClause && !SafeToPushdownWindowFunction(subquery,
&errorDetail))
{
return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED, errorDetail->data, NULL,
NULL);
}
if (subquery->setOperations != NULL)
{

109
src/backend/distributed/planner/multi_logical_planner.c
View File

@ -83,7 +83,7 @@ static DeferredErrorMessage * DeferErrorIfUnsupportedUnionQuery(Query *queryTree
outerMostQueryHasLimit);
static bool ExtractSetOperationStatmentWalker(Node *node, List **setOperationList);
static DeferredErrorMessage * DeferErrorIfUnsupportedTableCombination(Query *queryTree);
static bool TargetListOnPartitionColumn(Query *query, List *targetEntryList);
static bool WindowPartitionOnDistributionColumn(Query *query);
static FieldSelect * CompositeFieldRecursive(Expr *expression, Query *query);
static bool FullCompositeFieldList(List *compositeFieldList);
static MultiNode * MultiPlanTree(Query *queryTree);
@ -447,6 +447,7 @@ MultiSubqueryPlanTree(Query *originalQuery, Query *queryTree,
* - Only a single RTE_RELATION exists, which means only a single table
* name is specified on the whole query
* - No sublinks exists in the subquery
* - No window functions in the subquery
*
* Note that the caller should still call DeferErrorIfUnsupportedSubqueryRepartition()
* to ensure that Citus supports the subquery. Also, this function is designed to run
@ -466,6 +467,12 @@ SingleRelationRepartitionSubquery(Query *queryTree)
return false;
}
/* we don't support window functions */
if (queryTree->hasWindowFuncs)
{
return false;
}
/*
* Don't allow joins and set operations. If join appears in the queryTree, the
* length would be greater than 1. If only set operations exists, the length
@ -782,6 +789,7 @@ DeferErrorIfCannotPushdownSubquery(Query *subqueryTree, bool outerMostQueryHasLi
{
bool preconditionsSatisfied = true;
char *errorDetail = NULL;
StringInfo errorInfo = NULL;
DeferredErrorMessage *deferredError = NULL;
deferredError = DeferErrorIfUnsupportedTableCombination(subqueryTree);
@ -796,12 +804,6 @@ DeferErrorIfCannotPushdownSubquery(Query *subqueryTree, bool outerMostQueryHasLi
errorDetail = "Subqueries without relations are unsupported";
}
if (subqueryTree->hasWindowFuncs)
{
preconditionsSatisfied = false;
errorDetail = "Window functions are currently unsupported";
}
if (subqueryTree->limitOffset)
{
preconditionsSatisfied = false;
@ -871,6 +873,17 @@ DeferErrorIfCannotPushdownSubquery(Query *subqueryTree, bool outerMostQueryHasLi
}
}
/*
* We support window functions when the window function
* is partitioned on distribution column.
*/
if (subqueryTree->windowClause && !SafeToPushdownWindowFunction(subqueryTree,
&errorInfo))
{
errorDetail = (char *) errorInfo->data;
preconditionsSatisfied = false;
}
/* we don't support aggregates without group by */
if (subqueryTree->hasAggs && (subqueryTree->groupClause == NULL))
{
@ -1083,11 +1096,91 @@ DeferErrorIfUnsupportedTableCombination(Query *queryTree)
}
/*
* SafeToPushdownWindowFunction checks if the query with window function is supported.
* It returns the result accordingly and modifies the error detail.
*/
bool
SafeToPushdownWindowFunction(Query *query, StringInfo *errorDetail)
{
ListCell *windowClauseCell = NULL;
List *windowClauseList = query->windowClause;
/*
* We need to check each window clause separately if there is a partition by clause
* and if it is partitioned on the distribution column.
*/
foreach(windowClauseCell, windowClauseList)
{
WindowClause *windowClause = lfirst(windowClauseCell);
if (!windowClause->partitionClause)
{
*errorDetail = makeStringInfo();
appendStringInfoString(*errorDetail,
"Window functions without PARTITION BY on distribution "
"column is currently unsupported");
return false;
}
}
if (!WindowPartitionOnDistributionColumn(query))
{
*errorDetail = makeStringInfo();
appendStringInfoString(*errorDetail,
"Window functions with PARTITION BY list missing distribution "
"column is currently unsupported");
return false;
}
return true;
}
/*
* WindowPartitionOnDistributionColumn checks if the given subquery has one
* or more window functions and at least one of them is not partitioned by
* distribution column. The function returns false if your window function does not
* have a partition by clause or it does not include the distribution column.
*
* Please note that if the query does not have a window function, the function
* returns true.
*/
static bool
WindowPartitionOnDistributionColumn(Query *query)
{
List *windowClauseList = query->windowClause;
ListCell *windowClauseCell = NULL;
foreach(windowClauseCell, windowClauseList)
{
WindowClause *windowClause = lfirst(windowClauseCell);
List *groupTargetEntryList = NIL;
bool partitionOnDistributionColumn = false;
List *partitionClauseList = windowClause->partitionClause;
List *targetEntryList = query->targetList;
groupTargetEntryList =
GroupTargetEntryList(partitionClauseList, targetEntryList);
partitionOnDistributionColumn =
TargetListOnPartitionColumn(query, groupTargetEntryList);
if (!partitionOnDistributionColumn)
{
return false;
}
}
return true;
}
/*
* TargetListOnPartitionColumn checks if at least one target list entry is on
* partition column.
*/
static bool
bool
TargetListOnPartitionColumn(Query *query, List *targetEntryList)
{
bool targetListOnPartitionColumn = false;

2
src/include/distributed/multi_logical_planner.h
View File

@ -185,6 +185,8 @@ extern MultiTreeRoot * MultiLogicalPlanCreate(Query *originalQuery, Query *query
PlannerRestrictionContext *
plannerRestrictionContext,
ParamListInfo boundParams);
extern bool SafeToPushdownWindowFunction(Query *query, StringInfo *errorDetail);
extern bool TargetListOnPartitionColumn(Query *query, List *targetEntryList);
extern bool NeedsDistributedPlanning(Query *queryTree);
extern MultiNode * ParentNode(MultiNode *multiNode);
extern MultiNode * ChildNode(MultiUnaryNode *multiNode);

13
src/test/regress/expected/multi_insert_select.out
View File

@ -1015,8 +1015,7 @@ FROM (SELECT SUM(raw_events_second.value_4) AS v4,
raw_events_second
WHERE raw_events_first.user_id = raw_events_second.user_id
GROUP BY raw_events_second.value_3) AS foo;
DEBUG: cannot perform distributed INSERT INTO ... SELECT because the partition columns in the source table and subquery do not match
DETAIL: The data type of the target table's partition column should exactly match the data type of the corresponding simple column reference in the subquery.
DEBUG: Group by list without distribution column is not allowed in distributed INSERT ... SELECT queries
DEBUG: Collecting INSERT ... SELECT results on coordinator
ERROR: cannot push down this subquery
DETAIL: Group by list without partition column is currently unsupported
@ -1133,9 +1132,7 @@ FROM (SELECT SUM(raw_events_second.value_4) AS v4,
GROUP BY raw_events_second.value_1
HAVING SUM(raw_events_second.value_4) > 10) AS foo2 ) as f2
ON (f.id = f2.id);
DEBUG: cannot perform distributed INSERT INTO ... SELECT because the partition columns in the source table and subquery do not match
DETAIL: Subquery contains an expression that is not a simple column reference in the same position as the target table's partition column.
HINT: Ensure the target table's partition column has a corresponding simple column reference to a distributed table's partition column in the subquery.
DEBUG: Group by list without distribution column is not allowed in distributed INSERT ... SELECT queries
DEBUG: Collecting INSERT ... SELECT results on coordinator
ERROR: cannot pushdown the subquery since all relations are not joined using distribution keys
DETAIL: Each relation should be joined with at least one another relation using distribution keys and equality operator.
@ -1164,8 +1161,10 @@ FROM (SELECT SUM(raw_events_second.value_4) AS v4,
GROUP BY raw_events_second.value_1
HAVING SUM(raw_events_second.value_4) > 10) AS foo2 ) as f2
ON (f.id = f2.id);
ERROR: cannot perform distributed planning for the given modification
DETAIL: Select query cannot be pushed down to the worker.
DEBUG: Group by list without distribution column is not allowed in distributed INSERT ... SELECT queries
DEBUG: Collecting INSERT ... SELECT results on coordinator
ERROR: cannot pushdown the subquery since all relations are not joined using distribution keys
DETAIL: Each relation should be joined with at least one another relation using distribution keys and equality operator.
-- cannot pushdown the query since the JOIN is not equi JOIN
INSERT INTO agg_events
(user_id, value_4_agg)

855
src/test/regress/expected/multi_insert_select_window.out Normal file
View File

@ -0,0 +1,855 @@
-- ===================================================================
-- test insert select functionality for window functions
-- ===================================================================
TRUNCATE agg_results;
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
10001 | 101 | 49.5810418958104190
(1 row)
TRUNCATE agg_results;
-- the same test with different syntax
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY time DESC) as rnk
FROM
events_table
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
10001 | 101 | 49.5810418958104190
(1 row)
TRUNCATE agg_results;
-- similar test with lag
INSERT INTO agg_results (user_id, agg_time, value_2_agg, value_3_agg)
SELECT
user_id, time, lag_event_type, row_no
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type, row_number() OVER my_win as row_no
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
10001 | 101 | 49.5810418958104190
(1 row)
TRUNCATE agg_results;
-- simple window function, partitioned and grouped by on the distribution key
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT
user_id, rnk, tme
FROM
(
SELECT
user_id, rank() OVER my_win as rnk, avg(value_2) as tme
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY avg(event_type) DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
1188 | 101 | 49.7895622895622896
(1 row)
TRUNCATE agg_results;
-- top level query has a group by on the result of the window function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
min(user_id), min(time), lag_event_type
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo
GROUP BY
lag_event_type;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+--------------------
1002 | 50 | 9.7844311377245509
(1 row)
TRUNCATE agg_results;
-- window functions should work along with joins as well
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
195 | 91 | 51.0205128205128205
(1 row)
TRUNCATE agg_results;
-- two window functions in a single subquery should work fine as well
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
202 | 91 | 50.2970297029702970
(1 row)
TRUNCATE agg_results;
-- window functions should be fine within subquery joins
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg, value_3_agg)
SELECT sub_1.user_id, max(lag_1), max(rank_1), max(rank_2) FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_1, rank() OVER w2 as rank_1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as sub_1
JOIN
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_2, rank() OVER w2 as rank_2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.value_2 ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 50) ORDER BY events_table.time)
) as sub_2
ON(sub_1.user_id = sub_2.user_id)
GROUP BY
sub_1.user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
91 | 91 | 50.2637362637362637
(1 row)
TRUNCATE agg_results;
-- GROUP BYs and PARTITION BYs should work fine together
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 5
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
7 | 6 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- aggregates in the PARTITION BY is also allows
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id, avg(event_type%10)::int ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 0
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
8 | 7 | 48.8750000000000000
(1 row)
TRUNCATE agg_results;
-- GROUP BY should not necessarly be inclusive of partitioning
-- but this query doesn't make much sense
INSERT INTO agg_results (user_id, value_1_agg)
SELECT
avg(user_id), my_rank
FROM
(
SELECT
user_id, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id
WINDOW my_win AS (PARTITION BY user_id, max(event_type) ORDER BY count(*) DESC)
) as foo
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
1 | 1 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- Group by has more columns than partition by which uses coordinator insert ... select
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, SUM(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
ORDER BY
2 DESC, 1
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
10 | 10 | 49.1000000000000000
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_2_agg)
SELECT user_id, max(sum) FROM (
SELECT
user_id, SUM(value_2) OVER (PARTITION BY user_id, value_1)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
GROUP BY user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
101 | 101 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- Subquery in where with window function
INSERT INTO agg_results(user_id)
SELECT
user_id
FROM
users_table
WHERE
value_2 > 545 AND
value_2 < ALL (
SELECT
avg(value_3) OVER (PARTITION BY user_id)
FROM
events_table
WHERE
users_table.user_id = events_table.user_id
)
GROUP BY
user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
4 | 4 | 35.2500000000000000
(1 row)
TRUNCATE agg_results;
-- Partition by with aggregate functions. This query does not make much sense since the
-- result of aggregate function will be the same for every row in a partition and it is
-- not going to affect the group that the count function will work on.
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY user_id, MIN(value_2))
FROM
users_table
GROUP BY
1
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
101 | 101 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- Some more nested queries
INSERT INTO agg_results(user_id, value_2_agg, value_3_agg, value_4_agg)
SELECT
user_id, rank, SUM(ABS(value_2 - value_3)) AS difference, COUNT(*) AS distinct_users
FROM (
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY value_2 DESC)
FROM (
SELECT
user_id, value_2, sum(value_3) OVER (PARTITION BY user_id, value_2) as value_3
FROM users_table
) AS A
) AS A
GROUP BY
user_id, rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
9501 | 101 | 49.8461214608988528
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT * FROM (
SELECT DISTINCT
f3.user_id, ABS(f2.sum - f3.sum)
FROM (
SELECT DISTINCT
user_id, sum(value_3) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_3
) f3,
(
SELECT DISTINCT
user_id, sum(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_2
) f2
WHERE
f3.user_id=f2.user_id
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
101 | 101 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- test with reference table partitioned on columns from both
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
(
SELECT
DISTINCT user_id, count(id) OVER (PARTITION BY user_id, id)
FROM
users_table, users_ref_test_table
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
101 | 101 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- Window functions with HAVING clause
INSERT INTO agg_results (user_id, value_1_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1 HAVING count(*) > 1
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
437 | 100 | 49.9496567505720824
(1 row)
TRUNCATE agg_results;
-- Window functions with HAVING clause which uses coordinator insert ... select
INSERT INTO agg_results (user_id, value_1_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1 HAVING count(*) > 1
) a
ORDER BY
2 DESC, 1
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
10 | 5 | 32.4000000000000000
(1 row)
TRUNCATE agg_results;
-- Window function in View works
CREATE VIEW view_with_window_func AS
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1
HAVING count(*) > 1;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
view_with_window_func;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
437 | 100 | 49.9496567505720824
(1 row)
TRUNCATE agg_results;
-- Window function in View works and the query uses coordinator insert ... select
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
view_with_window_func
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot run avg(user_id)
SELECT count(*) FROM agg_results;
count
-------
10
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT
user_id, max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY
user_id
LIMIT
5;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot test avg or distinct count
SELECT count(*) FROM agg_results;
count
-------
5
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT
user_id, max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY
user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
94 | 94 | 50.4787234042553191
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop
LIMIT
5;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot test avg or distinct count
SELECT count(*) FROM agg_results;
count
-------
5
(1 row)
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
count | count | avg
-------+-------+---------------------
101 | 101 | 50.0000000000000000
(1 row)
TRUNCATE agg_results;
-- lets have some queries that Citus shouldn't push down
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY event_type ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
-- user needs to supply partition by which should
-- include the distribution key
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS ()
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
ERROR: cannot push down this subquery
DETAIL: Window functions without PARTITION BY on distribution column is currently unsupported
-- user needs to supply partition by which should
-- include the distribution key
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
ERROR: cannot push down this subquery
DETAIL: Window functions without PARTITION BY on distribution column is currently unsupported
-- w2 should not be pushed down
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id+1, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo
LIMIT
10;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
-- GROUP BY includes the partition key, but not the WINDOW function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125;
ERROR: cannot push down this subquery
DETAIL: Window functions without PARTITION BY on distribution column is currently unsupported
-- GROUP BY includes the partition key, but not the WINDOW function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY date_trunc('day', time) ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
-- w2 should not be allowed
INSERT INTO agg_results (user_id, value_2_agg, value_3_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (ORDER BY events_table.time)
) as foo;
ERROR: cannot push down this subquery
DETAIL: Window functions without PARTITION BY on distribution column is currently unsupported
-- unsupported window function with an override
INSERT INTO agg_results(user_id, agg_time, value_2_agg)
SELECT * FROM (
SELECT
user_id, date_trunc('day', time) as time, sum(rank) OVER w2
FROM (
SELECT DISTINCT
user_id as user_id, time, rank() over w1
FROM
users_table
WINDOW
w AS (PARTITION BY time), w1 AS (w ORDER BY value_2, value_3)
) fab
WINDOW
w2 as (PARTITION BY user_id, time)
) a;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
-- Subquery in where with unsupported window function
INSERT INTO agg_results(user_id)
SELECT
user_id
FROM
users_table
WHERE
value_2 > 545 AND
value_2 < ALL (
SELECT
avg(value_3) OVER ()
FROM
events_table
WHERE
users_table.user_id = events_table.user_id
)
GROUP BY
user_id;
ERROR: cannot push down this subquery
DETAIL: Window functions without PARTITION BY on distribution column is currently unsupported
-- Aggregate function on distribution column should error out
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY sum(user_id), MIN(value_2))
FROM
users_table
GROUP BY
user_id
) a;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
-- UNION with only one subquery which has a partition on non-distribution column should
-- error out
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by event_type) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop;
ERROR: cannot push down this subquery
DETAIL: Window functions with PARTITION BY list missing distribution column is currently unsupported
DROP VIEW view_with_window_func;

13
src/test/regress/expected/multi_subquery_complex_queries.out
View File

@ -2381,19 +2381,6 @@ ORDER BY
types;
ERROR: cannot push down this subquery
DETAIL: Offset clause is currently unsupported
-- not supported due to window functions
SELECT user_id,
some_vals
FROM (
SELECT * ,
Row_number() over (PARTITION BY "user_id" ORDER BY "user_id") AS "some_vals",
Random()
FROM users_table
) user_id
ORDER BY 1,
2 limit 10;
ERROR: cannot perform distributed planning on this query
DETAIL: Subqueries without group by clause are not supported yet
-- not supported due to non relation rte
SELECT ("final_query"."event_types") as types, count(*) AS sumOfEventType
FROM

14
src/test/regress/expected/multi_subquery_complex_reference_clause.out
View File

@ -6,7 +6,6 @@
-- We don't need shard id sequence here, so commented out to prevent conflicts with concurrent tests
-- ALTER SEQUENCE pg_catalog.pg_dist_shardid_seq RESTART 1400000;
ALTER SEQUENCE pg_catalog.pg_dist_jobid_seq RESTART 1400000;
SET citus.enable_router_execution TO FALSE;
CREATE TABLE user_buy_test_table(user_id int, item_id int, buy_count int);
SELECT create_distributed_table('user_buy_test_table', 'user_id');
@ -29,19 +28,6 @@ SELECT create_distributed_table('users_return_test_table', 'user_id');
INSERT INTO users_return_test_table VALUES(4,1,1);
INSERT INTO users_return_test_table VALUES(1,3,1);
INSERT INTO users_return_test_table VALUES(3,2,2);
CREATE TABLE users_ref_test_table(id int, it_name varchar(25), k_no int);
SELECT create_reference_table('users_ref_test_table');
create_reference_table
------------------------
(1 row)
INSERT INTO users_ref_test_table VALUES(1,'User_1',45);
INSERT INTO users_ref_test_table VALUES(2,'User_2',46);
INSERT INTO users_ref_test_table VALUES(3,'User_3',47);
INSERT INTO users_ref_test_table VALUES(4,'User_4',48);
INSERT INTO users_ref_test_table VALUES(5,'User_5',49);
INSERT INTO users_ref_test_table VALUES(6,'User_6',50);
-- Simple Join test with reference table
SELECT count(*) FROM
(SELECT random() FROM user_buy_test_table JOIN users_ref_test_table

1050
src/test/regress/expected/multi_subquery_window_functions.out Normal file
View File

File diff suppressed because it is too large Load Diff

11
src/test/regress/input/multi_behavioral_analytics_create_table.source
View File

@ -2,10 +2,8 @@
-- multi behavioral analytics
-- this file is intended to create the table requires for the tests
--
ALTER SEQUENCE pg_catalog.pg_dist_shardid_seq RESTART 1400000;
ALTER SEQUENCE pg_catalog.pg_dist_jobid_seq RESTART 1400000;
SET citus.shard_replication_factor = 1;
SET citus.shard_count = 4;
@ -30,6 +28,15 @@ SELECT create_distributed_table('agg_results_third', 'user_id');
CREATE TABLE agg_results_fourth (user_id int, value_1_agg int, value_2_agg int, value_3_agg float, value_4_agg bigint, agg_time timestamp);
SELECT create_distributed_table('agg_results_fourth', 'user_id');
CREATE TABLE users_ref_test_table(id int, it_name varchar(25), k_no int);
SELECT create_reference_table('users_ref_test_table');
INSERT INTO users_ref_test_table VALUES(1,'User_1',45);
INSERT INTO users_ref_test_table VALUES(2,'User_2',46);
INSERT INTO users_ref_test_table VALUES(3,'User_3',47);
INSERT INTO users_ref_test_table VALUES(4,'User_4',48);
INSERT INTO users_ref_test_table VALUES(5,'User_5',49);
INSERT INTO users_ref_test_table VALUES(6,'User_6',50);
COPY users_table FROM '@abs_srcdir@/data/users_table.data' WITH CSV;
COPY events_table FROM '@abs_srcdir@/data/events_table.data' WITH CSV;

4
src/test/regress/multi_schedule
View File

@ -36,7 +36,7 @@ test: multi_load_data
test: multi_behavioral_analytics_create_table
test: multi_behavioral_analytics_basics multi_behavioral_analytics_single_shard_queries multi_insert_select_non_pushable_queries
test: multi_insert_select
test: multi_insert_select multi_insert_select_window
# ---
# Tests for partitioning support
@ -51,7 +51,7 @@ test: multi_deparse_shard_query multi_distributed_transaction_id
test: multi_basic_queries multi_complex_expressions
test: multi_explain
test: multi_subquery multi_subquery_complex_queries multi_subquery_behavioral_analytics
test: multi_subquery_complex_reference_clause
test: multi_subquery_complex_reference_clause multi_subquery_window_functions
test: multi_subquery_in_where_reference_clause
test: multi_subquery_union multi_subquery_in_where_clause multi_subquery_misc
test: multi_reference_table

13
src/test/regress/output/multi_behavioral_analytics_create_table.source
View File

@ -51,6 +51,19 @@ SELECT create_distributed_table('agg_results_fourth', 'user_id');
(1 row)
CREATE TABLE users_ref_test_table(id int, it_name varchar(25), k_no int);
SELECT create_reference_table('users_ref_test_table');
create_reference_table
------------------------
(1 row)
INSERT INTO users_ref_test_table VALUES(1,'User_1',45);
INSERT INTO users_ref_test_table VALUES(2,'User_2',46);
INSERT INTO users_ref_test_table VALUES(3,'User_3',47);
INSERT INTO users_ref_test_table VALUES(4,'User_4',48);
INSERT INTO users_ref_test_table VALUES(5,'User_5',49);
INSERT INTO users_ref_test_table VALUES(6,'User_6',50);
COPY users_table FROM '@abs_srcdir@/data/users_table.data' WITH CSV;
COPY events_table FROM '@abs_srcdir@/data/events_table.data' WITH CSV;
-- create indexes for

769
src/test/regress/sql/multi_insert_select_window.sql Normal file
View File

@ -0,0 +1,769 @@
-- ===================================================================
-- test insert select functionality for window functions
-- ===================================================================
TRUNCATE agg_results;
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- the same test with different syntax
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY time DESC) as rnk
FROM
events_table
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- similar test with lag
INSERT INTO agg_results (user_id, agg_time, value_2_agg, value_3_agg)
SELECT
user_id, time, lag_event_type, row_no
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type, row_number() OVER my_win as row_no
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- simple window function, partitioned and grouped by on the distribution key
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT
user_id, rnk, tme
FROM
(
SELECT
user_id, rank() OVER my_win as rnk, avg(value_2) as tme
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY avg(event_type) DESC)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- top level query has a group by on the result of the window function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
min(user_id), min(time), lag_event_type
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo
GROUP BY
lag_event_type;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- window functions should work along with joins as well
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- two window functions in a single subquery should work fine as well
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- window functions should be fine within subquery joins
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg, value_3_agg)
SELECT sub_1.user_id, max(lag_1), max(rank_1), max(rank_2) FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_1, rank() OVER w2 as rank_1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as sub_1
JOIN
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_2, rank() OVER w2 as rank_2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.value_2 ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 50) ORDER BY events_table.time)
) as sub_2
ON(sub_1.user_id = sub_2.user_id)
GROUP BY
sub_1.user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- GROUP BYs and PARTITION BYs should work fine together
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 5
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- aggregates in the PARTITION BY is also allows
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id, avg(event_type%10)::int ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 0
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- GROUP BY should not necessarly be inclusive of partitioning
-- but this query doesn't make much sense
INSERT INTO agg_results (user_id, value_1_agg)
SELECT
avg(user_id), my_rank
FROM
(
SELECT
user_id, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id
WINDOW my_win AS (PARTITION BY user_id, max(event_type) ORDER BY count(*) DESC)
) as foo
GROUP BY
my_rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Group by has more columns than partition by which uses coordinator insert ... select
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, SUM(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
ORDER BY
2 DESC, 1
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_2_agg)
SELECT user_id, max(sum) FROM (
SELECT
user_id, SUM(value_2) OVER (PARTITION BY user_id, value_1)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
GROUP BY user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Subquery in where with window function
INSERT INTO agg_results(user_id)
SELECT
user_id
FROM
users_table
WHERE
value_2 > 545 AND
value_2 < ALL (
SELECT
avg(value_3) OVER (PARTITION BY user_id)
FROM
events_table
WHERE
users_table.user_id = events_table.user_id
)
GROUP BY
user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Partition by with aggregate functions. This query does not make much sense since the
-- result of aggregate function will be the same for every row in a partition and it is
-- not going to affect the group that the count function will work on.
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY user_id, MIN(value_2))
FROM
users_table
GROUP BY
1
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Some more nested queries
INSERT INTO agg_results(user_id, value_2_agg, value_3_agg, value_4_agg)
SELECT
user_id, rank, SUM(ABS(value_2 - value_3)) AS difference, COUNT(*) AS distinct_users
FROM (
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY value_2 DESC)
FROM (
SELECT
user_id, value_2, sum(value_3) OVER (PARTITION BY user_id, value_2) as value_3
FROM users_table
) AS A
) AS A
GROUP BY
user_id, rank;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT * FROM (
SELECT DISTINCT
f3.user_id, ABS(f2.sum - f3.sum)
FROM (
SELECT DISTINCT
user_id, sum(value_3) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_3
) f3,
(
SELECT DISTINCT
user_id, sum(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_2
) f2
WHERE
f3.user_id=f2.user_id
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- test with reference table partitioned on columns from both
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
(
SELECT
DISTINCT user_id, count(id) OVER (PARTITION BY user_id, id)
FROM
users_table, users_ref_test_table
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Window functions with HAVING clause
INSERT INTO agg_results (user_id, value_1_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1 HAVING count(*) > 1
) a;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Window functions with HAVING clause which uses coordinator insert ... select
INSERT INTO agg_results (user_id, value_1_agg)
SELECT * FROM (
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1 HAVING count(*) > 1
) a
ORDER BY
2 DESC, 1
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Window function in View works
CREATE VIEW view_with_window_func AS
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1
HAVING count(*) > 1;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
view_with_window_func;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- Window function in View works and the query uses coordinator insert ... select
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM
view_with_window_func
LIMIT
10;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot run avg(user_id)
SELECT count(*) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT
user_id, max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY
user_id
LIMIT
5;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot test avg or distinct count
SELECT count(*) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT
user_id, max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY
user_id;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop
LIMIT
5;
-- get some statistics from the aggregated results to ensure the results are correct
-- since there is a limit but not order, we cannot test avg or distinct count
SELECT count(*) FROM agg_results;
TRUNCATE agg_results;
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop;
-- get some statistics from the aggregated results to ensure the results are correct
SELECT count(*), count(DISTINCT user_id), avg(user_id) FROM agg_results;
TRUNCATE agg_results;
-- lets have some queries that Citus shouldn't push down
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY event_type ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- user needs to supply partition by which should
-- include the distribution key
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS ()
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- user needs to supply partition by which should
-- include the distribution key
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- w2 should not be pushed down
INSERT INTO agg_results (user_id, value_1_agg, value_2_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id+1, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo
LIMIT
10;
-- GROUP BY includes the partition key, but not the WINDOW function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125;
-- GROUP BY includes the partition key, but not the WINDOW function
INSERT INTO agg_results (user_id, agg_time, value_2_agg)
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY date_trunc('day', time) ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125;
-- w2 should not be allowed
INSERT INTO agg_results (user_id, value_2_agg, value_3_agg)
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (ORDER BY events_table.time)
) as foo;
-- unsupported window function with an override
INSERT INTO agg_results(user_id, agg_time, value_2_agg)
SELECT * FROM (
SELECT
user_id, date_trunc('day', time) as time, sum(rank) OVER w2
FROM (
SELECT DISTINCT
user_id as user_id, time, rank() over w1
FROM
users_table
WINDOW
w AS (PARTITION BY time), w1 AS (w ORDER BY value_2, value_3)
) fab
WINDOW
w2 as (PARTITION BY user_id, time)
) a;
-- Subquery in where with unsupported window function
INSERT INTO agg_results(user_id)
SELECT
user_id
FROM
users_table
WHERE
value_2 > 545 AND
value_2 < ALL (
SELECT
avg(value_3) OVER ()
FROM
events_table
WHERE
users_table.user_id = events_table.user_id
)
GROUP BY
user_id;
-- Aggregate function on distribution column should error out
INSERT INTO agg_results(user_id, value_2_agg)
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY sum(user_id), MIN(value_2))
FROM
users_table
GROUP BY
user_id
) a;
-- UNION with only one subquery which has a partition on non-distribution column should
-- error out
INSERT INTO agg_results(user_id, value_1_agg)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by event_type) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id
)
) AS ftop;
DROP VIEW view_with_window_func;

12
src/test/regress/sql/multi_subquery_complex_queries.sql
View File

@ -2160,18 +2160,6 @@ GROUP BY
ORDER BY
types;
-- not supported due to window functions
SELECT user_id,
some_vals
FROM (
SELECT * ,
Row_number() over (PARTITION BY "user_id" ORDER BY "user_id") AS "some_vals",
Random()
FROM users_table
) user_id
ORDER BY 1,
2 limit 10;
-- not supported due to non relation rte
SELECT ("final_query"."event_types") as types, count(*) AS sumOfEventType
FROM

9
src/test/regress/sql/multi_subquery_complex_reference_clause.sql
View File

@ -23,15 +23,6 @@ INSERT INTO users_return_test_table VALUES(4,1,1);
INSERT INTO users_return_test_table VALUES(1,3,1);
INSERT INTO users_return_test_table VALUES(3,2,2);
CREATE TABLE users_ref_test_table(id int, it_name varchar(25), k_no int);
SELECT create_reference_table('users_ref_test_table');
INSERT INTO users_ref_test_table VALUES(1,'User_1',45);
INSERT INTO users_ref_test_table VALUES(2,'User_2',46);
INSERT INTO users_ref_test_table VALUES(3,'User_3',47);
INSERT INTO users_ref_test_table VALUES(4,'User_4',48);
INSERT INTO users_ref_test_table VALUES(5,'User_5',49);
INSERT INTO users_ref_test_table VALUES(6,'User_6',50);
-- Simple Join test with reference table
SELECT count(*) FROM
(SELECT random() FROM user_buy_test_table JOIN users_ref_test_table

706
src/test/regress/sql/multi_subquery_window_functions.sql Normal file
View File

@ -0,0 +1,706 @@
-- ===================================================================
-- test multi subquery functionality for window functions
-- ===================================================================
CREATE VIEW subq AS
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1
HAVING count(*) > 1;
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- the same test with different syntax
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY time DESC) as rnk
FROM
events_table
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- similar test with lag
SELECT
user_id, time, lag_event_type, row_no
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type, row_number() OVER my_win as row_no
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo
ORDER BY
4 DESC, 3 DESC NULLS LAST, 1 DESC, 2 DESC
LIMIT
10;
-- simple window function, partitioned and grouped by on the distribution key
SELECT
user_id, rnk, avg_val_2
FROM
(
SELECT
user_id, rank() OVER my_win as rnk, avg(value_2) as avg_val_2
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY avg(event_type) DESC)
) as foo
ORDER BY
2 DESC, 1 DESC, 3 DESC
LIMIT
10;
-- top level query has a group by on the result of the window function
SELECT
min(user_id), min(time), lag_event_type, count(*)
FROM
(
SELECT
*, lag(event_type) OVER my_win as lag_event_type
FROM
events_table WINDOW my_win AS (PARTITION BY user_id ORDER BY time DESC)
) as foo
GROUP BY
lag_event_type
ORDER BY
3 DESC NULLS LAST, 1 DESC, 2 DESC
LIMIT
10;
-- window functions should work along with joins as well
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time)
) as foo
ORDER BY 3 DESC, 1 DESC, 2 DESC NULLS LAST
LIMIT 10;
-- two window functions in a single subquery should work fine as well
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo
ORDER BY 3 DESC, 1 DESC, 2 DESC NULLS LAST
LIMIT 10;
-- window functions should be fine within subquery joins
SELECT sub_1.user_id, max(lag_1), max(rank_1), max(rank_2) FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_1, rank() OVER w2 as rank_1
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 25) ORDER BY events_table.time)
) as sub_1
JOIN
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1 as lag_2, rank() OVER w2 as rank_2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.value_2 ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id, (events_table.value_2 % 50) ORDER BY events_table.time)
) as sub_2
ON(sub_1.user_id = sub_2.user_id)
GROUP BY
sub_1.user_id
ORDER BY 3 DESC, 4 DESC, 1 DESC, 2 DESC NULLS LAST
LIMIT 10;
-- GROUP BYs and PARTITION BYs should work fine together
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 5
GROUP BY
my_rank
ORDER BY
3 DESC, 1 DESC,2 DESC
LIMIT
10;
-- aggregates in the PARTITION BY is also allows
SELECT
avg(user_id), max(time), my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY user_id, avg(event_type%10)::int ORDER BY count(*) DESC)
) as foo
WHERE
my_rank > 0
GROUP BY
my_rank
ORDER BY
3 DESC, 1 DESC,2 DESC
LIMIT
10;
-- GROUP BY should not necessarly be inclusive of partitioning
-- but this query doesn't make much sense
SELECT
avg(user_id), my_rank
FROM
(
SELECT
user_id, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id
WINDOW my_win AS (PARTITION BY user_id, max(event_type) ORDER BY count(*) DESC)
) as foo
GROUP BY
my_rank
ORDER BY
2 DESC, 1 DESC
LIMIT
10;
-- Using previously defined supported window function on distribution key
SELECT * FROM (
SELECT
user_id, date_trunc('day', time) as time, sum(rank) OVER w2
FROM (
SELECT DISTINCT
user_id as user_id, time, rank() over w1
FROM users_table
WINDOW
w AS (PARTITION BY user_id),
w1 AS (w ORDER BY value_2, value_3)
) fab
WINDOW
w2 as (PARTITION BY user_id, time)
) a
ORDER BY
1, 2, 3 DESC
LIMIT
10;
-- test with reference table partitioned on columns from both
SELECT *
FROM
(
SELECT
DISTINCT user_id, it_name, count(id) OVER (PARTITION BY user_id, id)
FROM
users_table, users_ref_test_table
WHERE users_table.value_2=users_ref_test_table.k_no
) a
ORDER BY
1, 2, 3
LIMIT
20;
-- Group by has more columns than partition by
SELECT * FROM (
SELECT
DISTINCT user_id, SUM(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
ORDER BY
2 DESC, 1
LIMIT
10;
SELECT user_id, max(sum) FROM (
SELECT
user_id, SUM(value_2) OVER (PARTITION BY user_id, value_1)
FROM
users_table
GROUP BY
user_id, value_1, value_2
) a
GROUP BY user_id ORDER BY
2 DESC,1
LIMIT
10;
-- Window functions with HAVING clause
SELECT * FROM (
SELECT
DISTINCT user_id, rank() OVER (PARTITION BY user_id ORDER BY value_1)
FROM
users_table
GROUP BY
user_id, value_1 HAVING count(*) > 1
) a
ORDER BY
2 DESC, 1
LIMIT
10;
-- Window function in View works
SELECT *
FROM
subq
ORDER BY
2 DESC, 1
LIMIT
10;
-- Window functions with UNION/UNION ALL works
SELECT
max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY user_id
ORDER BY 1 DESC
LIMIT 5;
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id)) AS ftop
ORDER BY 2 DESC, 1 DESC
LIMIT 5;
-- Subquery in where with window function
SELECT
user_id
FROM
users_table
WHERE
value_2 > 545 AND
value_2 < ALL (
SELECT
avg(value_3) OVER (PARTITION BY user_id)
FROM
events_table
WHERE
users_table.user_id = events_table.user_id
)
GROUP BY
user_id
ORDER BY
user_id DESC
LIMIT
3;
-- Some more nested queries
SELECT
user_id, rank, SUM(ABS(value_2 - value_3)) AS difference, COUNT(*) AS distinct_users
FROM (
SELECT
*, rank() OVER (PARTITION BY user_id ORDER BY value_2 DESC)
FROM (
SELECT
user_id, value_2, sum(value_3) OVER (PARTITION BY user_id, value_2) as value_3
FROM users_table
) AS A
) AS A
GROUP BY
user_id, rank
ORDER BY
difference DESC, rank DESC
LIMIT 20;
SELECT * FROM (
SELECT DISTINCT
f3.user_id, ABS(f2.sum - f3.sum)
FROM (
SELECT DISTINCT
user_id, sum(value_3) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_3
) f3,
(
SELECT DISTINCT
user_id, sum(value_2) OVER (PARTITION BY user_id)
FROM
users_table
GROUP BY
user_id, value_2
) f2
WHERE
f3.user_id=f2.user_id
) a
ORDER BY
abs DESC
LIMIT 10;
-- Partition by with aggregate functions. This query does not make much sense since the
-- result of aggregate function will be the same for every row in a partition and it is
-- not going to affect the group that the count function will work on.
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY user_id, MIN(value_2))
FROM
users_table
GROUP BY
1
) a
ORDER BY
1 DESC
LIMIT
5;
EXPLAIN (COSTS FALSE, VERBOSE TRUE)
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id)) AS ftop
ORDER BY 2 DESC, 1 DESC
LIMIT 5;
-- lets have some queries that Citus shouldn't push down
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (PARTITION BY event_type ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- user needs to supply partition by which should
-- include the distribution key
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS ()
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- user needs to supply partition by which should
-- include the distribution key
SELECT
user_id, time, rnk
FROM
(
SELECT
*, rank() OVER my_win as rnk
FROM
events_table
WINDOW my_win AS (ORDER BY time DESC)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC
LIMIT
10;
-- w2 should not be pushed down
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (PARTITION BY users_table.user_id+1, (events_table.value_2 % 25) ORDER BY events_table.time)
) as foo
ORDER BY 3 DESC, 1 DESC, 2 DESC NULLS LAST
LIMIT 10;
-- w2 should not be pushed down
SELECT * FROM
(
SELECT
DISTINCT users_table.user_id, lag(users_table.user_id) OVER w1, rank() OVER w2
FROM
users_table, events_table
WHERE
users_table.user_id = events_table.user_id and
event_type < 25
WINDOW w1 AS (PARTITION BY users_table.user_id, events_table.event_type ORDER BY events_table.time),
w2 AS (ORDER BY events_table.time)
) as foo
ORDER BY
3 DESC, 1 DESC, 2 DESC NULLS LAST
LIMIT
10;
-- GROUP BY includes the partition key, but not the WINDOW function
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125
ORDER BY
3 DESC, 1 DESC,2 DESC
LIMIT
10;
-- GROUP BY includes the partition key, but not the WINDOW function
SELECT
user_id, time, my_rank
FROM
(
SELECT
user_id, date_trunc('day', time) as time, rank() OVER my_win as my_rank
FROM
events_table
GROUP BY
user_id, date_trunc('day', time)
WINDOW my_win AS (PARTITION BY date_trunc('day', time) ORDER BY avg(event_type))
) as foo
WHERE
my_rank > 125
ORDER BY
3 DESC, 1 DESC,2 DESC
LIMIT
10;
-- Overriding window function but not supported
SELECT * FROM (
SELECT
user_id, date_trunc('day', time) as time, sum(rank) OVER w2
FROM (
SELECT DISTINCT
user_id as user_id, time, rank() over w1
FROM
users_table
WINDOW
w AS (PARTITION BY time), w1 AS (w ORDER BY value_2, value_3)
) fab
WINDOW
w2 as (PARTITION BY user_id, time)
) a
ORDER BY
1,2,3;
-- Aggregate function on distribution column should error out
SELECT * FROM (
SELECT
user_id, COUNT(*) OVER (PARTITION BY sum(user_id), MIN(value_2))
FROM
users_table
GROUP BY
user_id
) a
ORDER BY
1 DESC, 2 DESC;
-- test with reference table partitioned on only a column from reference table
SELECT *
FROM
(
SELECT
DISTINCT user_id, it_name, count(id) OVER (PARTITION BY id)
FROM
users_table, users_ref_test_table
) a
ORDER BY
1, 2, 3
LIMIT
20;
-- UNION ALL with only one of them is not partitioned over distribution column which
-- should not be allowed.
SELECT
max(avg)
FROM
(
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (1, 2, 3, 4, 5))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (6, 7, 8, 9, 10))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (11, 12, 13, 14, 15))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (16, 17, 18, 19, 20))
UNION ALL
(SELECT avg(value_3) over (partition by user_id), user_id FROM events_table where event_type IN (21, 22, 23, 24, 25))
UNION ALL
(SELECT avg(value_3) over (partition by event_type), user_id FROM events_table where event_type IN (26, 27, 28, 29, 30))
) b
GROUP BY user_id
ORDER BY 1 DESC
LIMIT 5;
-- UNION with only one subquery which has a partition on non-distribution column should
-- error out
SELECT *
FROM (
( SELECT user_id,
sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
events_table) user_id_1
GROUP BY
user_id)
UNION
(SELECT
user_id, sum(counter)
FROM
(SELECT
user_id, sum(value_2) over (partition by user_id) AS counter
FROM
users_table
UNION
SELECT
user_id, sum(value_2) over (partition by event_type) AS counter
FROM
events_table) user_id_2
GROUP BY
user_id)) AS ftop
ORDER BY 2 DESC, 1 DESC
LIMIT 5;
DROP VIEW subq;