hutiefang76 opened a new issue, #12542:
URL: https://github.com/apache/seatunnel/issues/12542
### Search before asking
I checked existing DuckDB issues and open PRs. #11485 addresses generic
timezone support for other dialects; this report concerns DuckDB's catalog
query path.
### What happened
DuckDB JDBC Source schema discovery for `query` uses the generic JDBC
converter instead of DuckDB's native mapper. With duckdb_jdbc 1.3.1.0:
| Native query type | Current inferred result | Expected |
| --- | --- | --- |
| DECIMAL(20,0) | BIGINT | DECIMAL(20,0), preserving precision and scale |
| TIMESTAMP WITH TIME ZONE | TIMESTAMP | TIMESTAMP_TZ |
| UUID / JSON / INTERVAL / HUGEINT | Unsupported JDBC type 1111 | Existing
DuckDB converter mapping |
| INTEGER[] / STRUCT | Unsupported JDBC type 2003 / 2002 | STRING,
consistent with existing native-type handling |
The same problems occur in ordinary DuckDB without DuckLake. Query aliases
and expressions need the native metadata mapping too. `table_path` already uses
the DuckDB converter; this report does not propose changing its discovery
behavior.
### SeaTunnel Version
3.0.0-SNAPSHOT, dev base 7fa3412e0; duckdb_jdbc 1.3.1.0.
### SeaTunnel Config
Create a local DuckDB table using JDBC:
```sql
CREATE TABLE query_types (id INTEGER, d DECIMAL(20,0), tz TIMESTAMPTZ, u
UUID);
INSERT INTO query_types VALUES (1,12345678901234567890,
TIMESTAMPTZ '2024-01-01 12:34:56.123456+08',
UUID '550e8400-e29b-41d4-a716-446655440000');
```
Then infer/read using this Source configuration:
```hocon
env { job.mode = "BATCH", parallelism = 1 }
source {
Jdbc {
url = "jdbc:duckdb:/tmp/query-types.db"
driver = "org.duckdb.DuckDBDriver"
query = "SELECT id,d,tz,u FROM query_types"
}
}
sink { Console {} }
```
Query `d` and `tz` separately to see the incorrect inferred types; including
`u` fails discovery.
### Running Command
Reproduced through `DuckDBCatalog.getTable(String)` and the actual
`JdbcSourceFactory`/Source flow, independent of the execution engine. The
companion PR extends the existing DuckDB catalog and Source/Sink test classes.
On the original production code, the four added regressions fail (two schema
assertions and two unsupported-type errors).
### Error Exception
```text
expected: Decimal(20, 0), actual: BIGINT
expected: TIMESTAMP_TZ, actual: TIMESTAMP
java.lang.UnsupportedOperationException: Unsupported JDBC type: 1111
```
### Java or Scala Version
JDK 8 and 17.
### Are you willing to submit PR?
Yes, a focused native query-metadata fix with regressions and EN/ZH
migration notes is ready.
### Code of Conduct
I agree to follow the project's Code of Conduct.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
