# name: test/sql/window/test_naive_aggregation.test # description: Test naive aggregation implementation # group: [window] statement ok SET default_null_order='nulls_first'; statement ok PRAGMA enable_verification statement ok PRAGMA debug_window_mode=separate statement ok CREATE TABLE empsalary (depname varchar, empno bigint, salary int, enroll_date date) statement ok INSERT INTO empsalary VALUES ('develop', 10, 5200, '2007-08-01'), ('sales', 1, 5000, '2006-10-01'), ('personnel', 5, 3500, '2007-12-10'), ('sales', 4, 4800, '2007-08-08'), ('personnel', 2, 3900, '2006-12-23'), ('develop', 7, 4200, '2008-01-01'), ('develop', 9, 4500, '2008-01-01'), ('sales', 3, 4800, '2007-08-01'), ('develop', 8, 6000, '2006-10-01'), ('develop', 11, 5200, '2007-08-15') # basic example from postgres' window.sql query TIIR SELECT depname, empno, salary, sum(salary) OVER (PARTITION BY depname ORDER BY empno) FROM empsalary ORDER BY depname, empno ---- develop 7 4200 4200.000000 develop 8 6000 10200.000000 develop 9 4500 14700.000000 develop 10 5200 19900.000000 develop 11 5200 25100.000000 personnel 2 3900 3900.000000 personnel 5 3500 7400.000000 sales 1 5000 5000.000000 sales 3 4800 9800.000000 sales 4 4800 14600.000000 # sum query R SELECT sum(salary) OVER (PARTITION BY depname ORDER BY salary) ss FROM empsalary ORDER BY depname, ss ---- 4200.000000 8700.000000 19100.000000 19100.000000 25100.000000 3500.000000 7400.000000 9600.000000 9600.000000 14600.000000 # min/max/avg query TIIR SELECT depname, min(salary) OVER (PARTITION BY depname ORDER BY salary, empno) m1, max(salary) OVER (PARTITION BY depname ORDER BY salary, empno) m2, AVG(salary) OVER (PARTITION BY depname ORDER BY salary, empno) m3 FROM empsalary ORDER BY depname, empno ---- develop 4200 4200 4200.000000 develop 4200 6000 5020.000000 develop 4200 4500 4350.000000 develop 4200 5200 4633.333333 develop 4200 5200 4775.000000 personnel 3500 3900 3700.000000 personnel 3500 3500 3500.000000 sales 4800 5000 4866.666667 sales 4800 4800 4800.000000 sales 4800 4800 4800.000000 # stddev_pop query TR SELECT depname, STDDEV_POP(salary) OVER (PARTITION BY depname ORDER BY salary, empno) s FROM empsalary ORDER BY depname, empno ---- develop 0.000000 develop 627.375486 develop 150.000000 develop 418.993503 develop 438.035387 personnel 200.000000 personnel 0.000000 sales 94.280904 sales 0.000000 sales 0.000000 # covar_pop query TR SELECT depname, COVAR_POP(salary, empno) OVER (PARTITION BY depname ORDER BY salary, empno) c FROM empsalary ORDER BY depname, empno ---- develop 0.000000 develop 240.000000 develop 150.000000 develop 477.777778 develop 606.250000 personnel -300.000000 personnel 0.000000 sales -111.111111 sales 0.000000 sales 0.000000 statement ok CREATE TABLE filtering AS SELECT x ,round(x * 0.333,0) % 3 AS y ,round(x * 0.333,0) % 3 AS z FROM generate_series(0,10) tbl(x); # The x_filtered_window and z_filtered_window columns should return a different output than plain_window query IIIIII SELECT x ,y ,z ,avg(x) OVER (PARTITION BY y) AS plain_window ,avg(x) FILTER (WHERE x = 1) OVER (PARTITION BY y) AS x_filtered_window ,avg(x) FILTER (WHERE z = 0) OVER (PARTITION BY y) AS z_filtered_window FROM filtering ORDER BY y, x; ---- 0 0.000000 0.000000 5.600000 1.000000 5.600000 1 0.000000 0.000000 5.600000 1.000000 5.600000 8 0.000000 0.000000 5.600000 1.000000 5.600000 9 0.000000 0.000000 5.600000 1.000000 5.600000 10 0.000000 0.000000 5.600000 1.000000 5.600000 2 1.000000 1.000000 3.000000 NULL NULL 3 1.000000 1.000000 3.000000 NULL NULL 4 1.000000 1.000000 3.000000 NULL NULL 5 2.000000 2.000000 6.000000 NULL NULL 6 2.000000 2.000000 6.000000 NULL NULL 7 2.000000 2.000000 6.000000 NULL NULL # COUNT(*) coverage query IIIIII SELECT x ,y ,z ,count(*) OVER (PARTITION BY y) AS plain_window ,count(*) FILTER (WHERE x = 1) OVER (PARTITION BY y) AS x_filtered_window ,count(*) FILTER (WHERE z = 0) OVER (PARTITION BY y) AS z_filtered_window FROM filtering ORDER BY y, x; ---- 0 0.000000 0.000000 5 1 5 1 0.000000 0.000000 5 1 5 8 0.000000 0.000000 5 1 5 9 0.000000 0.000000 5 1 5 10 0.000000 0.000000 5 1 5 2 1.000000 1.000000 3 0 0 3 1.000000 1.000000 3 0 0 4 1.000000 1.000000 3 0 0 5 2.000000 2.000000 3 0 0 6 2.000000 2.000000 3 0 0 7 2.000000 2.000000 3 0 0 # Holistic coverage query IIIIII SELECT x ,y ,z ,median(x) OVER (PARTITION BY y) AS plain_window ,median(x) FILTER (WHERE x = 1) OVER (PARTITION BY y) AS x_filtered_window ,median(x) FILTER (WHERE z = 0) OVER (PARTITION BY y) AS z_filtered_window FROM filtering ORDER BY y, x; ---- 0 0.000000 0.000000 8.000000 1.000000 8.000000 1 0.000000 0.000000 8.000000 1.000000 8.000000 8 0.000000 0.000000 8.000000 1.000000 8.000000 9 0.000000 0.000000 8.000000 1.000000 8.000000 10 0.000000 0.000000 8.000000 1.000000 8.000000 2 1.000000 1.000000 3.000000 NULL NULL 3 1.000000 1.000000 3.000000 NULL NULL 4 1.000000 1.000000 3.000000 NULL NULL 5 2.000000 2.000000 6.000000 NULL NULL 6 2.000000 2.000000 6.000000 NULL NULL 7 2.000000 2.000000 6.000000 NULL NULL # Filters do not affect framing. query II SELECT x, count(x) FILTER (WHERE x % 2 = 0) OVER (ORDER BY x ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING) FROM generate_series(0,10) tbl(x); ---- 0 2 1 2 2 3 3 2 4 3 5 2 6 3 7 2 8 3 9 2 10 2 # Test distinct hash table statement ok CREATE TABLE figure1 AS SELECT * FROM VALUES (1, 'a'), (2, 'b'), (3, 'b'), (4, 'c'), (5, 'c'), (6, 'b'), (7, 'c'), (8, 'a') v(i, s); query III SELECT i , s , COUNT(DISTINCT s) OVER( ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING) AS c FROM figure1 ORDER BY i ---- 1 a 2 2 b 3 3 b 3 4 c 2 5 c 2 6 b 3 7 c 3 8 a 3 # Test distinct and exclude query III SELECT i , s , COUNT(DISTINCT s) OVER( ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING EXCLUDE TIES) AS c FROM figure1 ORDER BY i ---- 1 a 2 2 b 3 3 b 3 4 c 2 5 c 2 6 b 3 7 c 3 8 a 3 # Test ORDER BY arguments # Multiple partitions => multiple threads. query III SELECT i // 10 AS p, i, ANY_VALUE(i ORDER BY i DESC) OVER( PARTITION BY i // 10 ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING ) AS c FROM range(20) tbl(i) ORDER BY ALL ---- 0 0 2 0 1 3 0 2 4 0 3 5 0 4 6 0 5 7 0 6 8 0 7 9 0 8 9 0 9 9 1 10 12 1 11 13 1 12 14 1 13 15 1 14 16 1 15 17 1 16 18 1 17 19 1 18 19 1 19 19 query III SELECT i // 10 AS p, i, LIST(i ORDER BY i DESC) OVER( PARTITION BY i // 10 ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING ) AS c FROM range(20) tbl(i) ORDER BY ALL ---- 0 0 [2, 1, 0] 0 1 [3, 2, 1, 0] 0 2 [4, 3, 2, 1, 0] 0 3 [5, 4, 3, 2, 1] 0 4 [6, 5, 4, 3, 2] 0 5 [7, 6, 5, 4, 3] 0 6 [8, 7, 6, 5, 4] 0 7 [9, 8, 7, 6, 5] 0 8 [9, 8, 7, 6] 0 9 [9, 8, 7] 1 10 [12, 11, 10] 1 11 [13, 12, 11, 10] 1 12 [14, 13, 12, 11, 10] 1 13 [15, 14, 13, 12, 11] 1 14 [16, 15, 14, 13, 12] 1 15 [17, 16, 15, 14, 13] 1 16 [18, 17, 16, 15, 14] 1 17 [19, 18, 17, 16, 15] 1 18 [19, 18, 17, 16] 1 19 [19, 18, 17] # Test DISTINCT + ORDER BY arguments query III SELECT i // 10 AS p, i, LIST(DISTINCT i // 2 ORDER BY i DESC) OVER( PARTITION BY i // 10 ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING ) AS c FROM range(20) tbl(i) ORDER BY ALL ---- 0 0 [1, 0] 0 1 [1, 0] 0 2 [2, 1, 0] 0 3 [2, 1, 0] 0 4 [3, 2, 1] 0 5 [3, 2, 1] 0 6 [4, 3, 2] 0 7 [4, 3, 2] 0 8 [4, 3] 0 9 [4, 3] 1 10 [6, 5] 1 11 [6, 5] 1 12 [7, 6, 5] 1 13 [7, 6, 5] 1 14 [8, 7, 6] 1 15 [8, 7, 6] 1 16 [9, 8, 7] 1 17 [9, 8, 7] 1 18 [9, 8] 1 19 [9, 8] # Test DISTINCT + ORDER BY + FILTER arguments query III SELECT i // 10 AS p, i, LIST(DISTINCT i // 2 ORDER BY i DESC) FILTER (i > 1) OVER( PARTITION BY i // 10 ORDER BY i ROWS BETWEEN 2 PRECEDING AND 2 FOLLOWING ) AS c FROM range(20) tbl(i) ORDER BY ALL ---- 0 0 [1] 0 1 [1] 0 2 [2, 1] 0 3 [2, 1] 0 4 [3, 2, 1] 0 5 [3, 2, 1] 0 6 [4, 3, 2] 0 7 [4, 3, 2] 0 8 [4, 3] 0 9 [4, 3] 1 10 [6, 5] 1 11 [6, 5] 1 12 [7, 6, 5] 1 13 [7, 6, 5] 1 14 [8, 7, 6] 1 15 [8, 7, 6] 1 16 [9, 8, 7] 1 17 [9, 8, 7] 1 18 [9, 8] 1 19 [9, 8]