# name: test/sql/topn/top_n_materialization.test
# description: Test Top N Optimization
# group: [topn]

statement ok
PRAGMA enable_verification

statement ok
CREATE TABLE test AS SELECT i, i + 1 AS j, i + 2 AS k, -i AS l FROM range(10) t(i);

statement ok
SET explain_output='optimized_only'

# Top N optimization
# this gets turned into a row-id join
query II
explain SELECT * FROM test ORDER BY j DESC LIMIT 2;
----
logical_opt	<REGEX>:.*COMPARISON_JOIN.*

query IIII
SELECT * FROM test ORDER BY j DESC LIMIT 2;
----
9	10	11	-9
8	9	10	-8

query II
explain SELECT * FROM test ORDER BY j, i LIMIT 2;
----
logical_opt	<REGEX>:.*COMPARISON_JOIN.*

query IIII
SELECT * FROM test ORDER BY j, i LIMIT 2;
----
0	1	2	0
1	2	3	-1

# this does not, we cannot remove any columns by turning it into a row-id join
query II
explain SELECT i FROM test ORDER BY i LIMIT 2;
----
logical_opt	<!REGEX>:.*COMPARISON_JOIN.*

# we cannot do this with volatile expressions
query II
explain SELECT * FROM (SELECT i + random() AS i, j, k, l FROM test) ORDER BY i LIMIT 2;
----
logical_opt	<!REGEX>:.*COMPARISON_JOIN.*

# top-n with expressions
query IIII
SELECT * FROM (SELECT -i i, -j j, -k k, -l l FROM test) ORDER BY -j DESC LIMIT 2
----
-9	-10	-11	9
-8	-9	-10	8

# multiple layers
query IIII
SELECT * FROM (SELECT 100 + i i, 1000 + j j, 10000 + k k, 100000 + l l FROM (SELECT -i i, -j j, -k k, -l l FROM test)) ORDER BY j DESC LIMIT 2
----
100	999	9998	100000
99	998	9997	100001

# limit + offset
query II
explain SELECT * FROM test LIMIT 2 OFFSET 2;
----
logical_opt	<REGEX>:.*COMPARISON_JOIN.*

query IIII
SELECT * FROM test LIMIT 2 OFFSET 2;
----
2	3	4	-2
3	4	5	-3

# sample
query II
explain SELECT * FROM test USING SAMPLE 2 ROWS
----
logical_opt	<REGEX>:.*COMPARISON_JOIN.*

# we can only use joins when we are sampling rows
query II
explain SELECT * FROM test USING SAMPLE 10%
----
logical_opt	<!REGEX>:.*COMPARISON_JOIN.*

# order on expression
query IIII
SELECT * FROM test ORDER BY -j DESC LIMIT 2
----
0	1	2	0
1	2	3	-1

# projection in subquery
query IIII
SELECT * FROM (SELECT -i i, -j j, -k k, -l l FROM test) ORDER BY -j DESC LIMIT 2
----
-9	-10	-11	9
-8	-9	-10	8

# filter after top-n
query IIII
SELECT * FROM (
	SELECT * FROM test ORDER BY j DESC LIMIT 2
) WHERE i=8
----
8	9	10	-8

query I
SELECT l FROM (
	SELECT * FROM test ORDER BY j DESC LIMIT 2
) WHERE k=10
----
-8

# now with varchar columns
statement ok
CREATE OR REPLACE TABLE test AS SELECT i, printf('%02d', i + 1) AS j, printf('%02d', i + 2) AS k, -i AS l FROM range(10) t(i);

query IIII
SELECT * FROM test ORDER BY j DESC LIMIT 2;
----
9	10	11	-9
8	09	10	-8

query IIII
SELECT j, k, l, i FROM test WHERE i > 5 ORDER BY j DESC LIMIT 2;
----
10	11	-9	9
09	10	-8	8

# test late materialization setting
statement ok
SET late_materialization_max_rows=0

query II
explain SELECT * FROM test ORDER BY j DESC LIMIT 2;
----
logical_opt	<!REGEX>:.*COMPARISON_JOIN.*

statement ok
RESET late_materialization_max_rows

query II
explain SELECT * FROM test ORDER BY j DESC LIMIT 2;
----
logical_opt	<REGEX>:.*COMPARISON_JOIN.*