Files
greptimedb/tests/cases/standalone/common/cte/cte.result
dennis zhuang bb9b7e8778 fix: re-scan stream-backed tables in recursive CTEs (#9039)
* fix: re-scan stream-backed tables in recursive CTEs

A recursive CTE re-executes its recursive term on every iteration, but
DfTableProviderAdapter hands StreamScanAdapter a single-use stream built at
planning time. The second iteration failed with "Stream already exhausted"
for every table served through DataSource::get_stream — information_schema,
pg_catalog, the computed entity-graph tables and numbers.

Keep that stream for the first execution and open a new one over the same
scan request for later executions.

Closes #9037

Signed-off-by: Dennis Zhuang <killme2008@gmail.com>

* refactor: drop redundant binding in stream factory

Signed-off-by: Dennis Zhuang <killme2008@gmail.com>

---------

Signed-off-by: Dennis Zhuang <killme2008@gmail.com>
2026-09-07 07:42:41 +00:00

194 lines
4.0 KiB
Plaintext

create table a(i integer, ts TIMESTAMP TIME INDEX);
Affected Rows: 0
insert into a values (42, 1);
Affected Rows: 1
with cte1 as (Select i as j from a) select * from cte1;
+----+
| j |
+----+
| 42 |
+----+
with cte1 as (Select i as j from a) select x from cte1 t1(x);
+----+
| x |
+----+
| 42 |
+----+
with cte1(xxx) as (Select i as j from a) select xxx from cte1;
+-----+
| xxx |
+-----+
| 42 |
+-----+
with cte1(xxx) as (Select i as j from a) select x from cte1 t1(x);
+----+
| x |
+----+
| 42 |
+----+
with cte1 as (Select i as j from a), cte2 as (select ref.j as k from cte1 as ref), cte3 as (select ref2.j+1 as i from cte1 as ref2) select * from cte2 , cte3;
+----+----+
| k | i |
+----+----+
| 42 | 43 |
+----+----+
-- SQLNESS SORT_RESULT 3 1
with cte1 as (select i as j from a), cte2 as (select ref.j as k from cte1 as ref), cte3 as (select ref2.j+1 as i from cte1 as ref2) select * from cte2 union all select * FROM cte3 order by 1;
+----+
| k |
+----+
| 42 |
| 43 |
+----+
with cte1 as (select 42), cte1 as (select 42) select * FROM cte1;
Error: 3000(PlanQuery), Failed to plan SQL: Error during planning: WITH query name "cte1" specified more than once
-- reference to CTE before its actually defined, it's not supported by datafusion
with cte3 as (select ref2.j as i from cte1 as ref2), cte1 as (Select i as j from a), cte2 as (select ref.j+1 as k from cte1 as ref) select * from cte2 union all select * FROM cte3;
Error: 4001(TableNotFound), Failed to plan SQL: Table not found: greptime.public.cte1
with cte1 as (Select i as j from a) select * from cte1 cte11, cte1 cte12;
+----+----+
| j | j |
+----+----+
| 42 | 42 |
+----+----+
with cte1 as (Select i as j from a) select * from cte1 where j = (select max(j) from cte1 as cte2);
+----+
| j |
+----+
| 42 |
+----+
with cte1(x, y) as (select 42 a, 84 b) select zzz, y from cte1 t1(zzz, y);
+-----+----+
| zzz | y |
+-----+----+
| 42 | 84 |
+-----+----+
SELECT 1 UNION ALL (WITH cte AS (SELECT 42) SELECT * FROM cte) order by 1;
+----------+
| Int64(1) |
+----------+
| 1 |
| 42 |
+----------+
WITH RECURSIVE cte(d) AS (
SELECT 1
UNION ALL
(WITH c(d) AS (SELECT * FROM cte)
SELECT d + 1
FROM c
WHERE FALSE
)
)
SELECT max(d) FROM cte;
+------------+
| max(cte.d) |
+------------+
| 1 |
+------------+
-- the recursive term is re-executed once per iteration, so its scans must be re-runnable
WITH RECURSIVE t AS (
SELECT 0 AS depth
UNION ALL
SELECT t.depth + 1 AS depth
FROM t, information_schema.tables
WHERE table_catalog = 'greptime'
AND table_schema = 'information_schema'
AND table_name = 'tables'
AND t.depth < 2
)
SELECT depth FROM t ORDER BY depth;
+-------+
| depth |
+-------+
| 0 |
| 1 |
| 2 |
+-------+
-- Nested aliases is not supported in datafusion
with cte (a) as (
select 1
)
select
a as alias1,
alias1 as alias2
from cte
where alias2 > 0;
Error: 3000(PlanQuery), Failed to plan SQL: No field named alias2. Valid fields are cte.a.
drop table a;
Affected Rows: 0
CREATE TABLE grpc_latencies
(
ts TIMESTAMP TIME INDEX,
host VARCHAR(255),
latency FLOAT,
PRIMARY KEY (host),
);
Affected Rows: 0
INSERT INTO grpc_latencies
VALUES ('2023-10-01 10:00:00', 'host1', 120),
('2023-10-01 10:00:00', 'host2', 150),
('2023-10-01 10:00:05', 'host1', 130);
Affected Rows: 3
WITH latencies AS (SELECT ts,
host,
AVG(latency) RANGE '2s' AS avg_latency
FROM grpc_latencies ALIGN '2s' BY (host) FILL PREV
)
SELECT latencies.ts,
AVG(latencies.avg_latency)
FROM latencies
GROUP BY latencies.ts ORDER BY latencies.ts;
+---------------------+----------------------------+
| ts | avg(latencies.avg_latency) |
+---------------------+----------------------------+
| 2023-10-01T10:00:00 | 135.0 |
| 2023-10-01T10:00:02 | 120.0 |
| 2023-10-01T10:00:04 | 130.0 |
+---------------------+----------------------------+
DROP TABLE grpc_latencies;
Affected Rows: 0