Repository navigation
MergeTree, filter: regexp with '?', negation with 'not' produce wrong results #92492
Description
Company or project name
No response
Describe what's wrong
Expecting these queries produce the same results, but they don't:
- using MergeTree as a source:
create or replace table in1 engine = MergeTree() order by url as
select 'http://example1.com/' as url;
-- expected: 1; returned: 0 - WRONG
select count(*) from in1
where not match(url, '^https?://clickhouse[.]com/');
- using CTE as a source:
-- same inputs, same regexp, same query, but without MergeTree
-- expected: 1 row; returned: 1 row - CORRECT
with
in1 as (
select 'http://example1.com/' as url
)
select count(*)
from in1
where not match(url, '^https?://clickhouse[.]com/');
More tests are here: https://fiddle.clickhouse.com/380c4fee-d7e4-4857-8eef-2b6b373a55b0
It seems that the presence of '?' inside regexp impacts the results.
Or negation using 'not', because where match() = false works fine.
Does it reproduce on the most recent release?
Yes
How to reproduce
https://fiddle.clickhouse.com/380c4fee-d7e4-4857-8eef-2b6b373a55b0
Expected behavior
No response
Error message and/or stacktrace
No response
Additional context
No response
Activity
KochetovNicolai commentedon Dec 19, 2025
It's a bug with KeyCondition.
If I replace
create or replace table in1 engine = MergeTree() order by urltocreate or replace table in1 engine = MergeTree() order by tuple()it works fine