ClickHouse/tests/queries/0_stateless/01748_partition_id_pruning.sql
2024-07-13 22:38:28 +00:00

36 lines
1.1 KiB
SQL

SET merge_tree_read_split_ranges_into_intersecting_and_non_intersecting_injection_probability = 0.0;
drop table if exists x;
create table x (i int, j int) engine MergeTree partition by i order by j settings index_granularity = 1;
insert into x values (1, 1), (1, 2), (1, 3), (2, 4), (2, 5), (2, 6);
set max_rows_to_read = 3;
select * from x where _partition_id = partitionID(1);
set max_rows_to_read = 5; -- one row for subquery + subquery
select * from x where _partition_id in (select partitionID(number + 1) from numbers(1));
-- trivial count optimization test
set max_rows_to_read = 2; -- one row for subquery + subquery itself
-- TODO: Relax the limits because we might build prepared set twice with _minmax_count_projection
set max_rows_to_read = 3;
select count() from x where _partition_id in (select partitionID(number + 1) from numbers(1));
drop table x;
drop table if exists mt;
create table mt (n UInt64) engine=MergeTree order by n partition by n % 10;
set max_rows_to_read = 200;
insert into mt select * from numbers(100);
select * from mt where toUInt64(substr(_part, 1, position(_part, '_') - 1)) = 1;
drop table mt;