tagadearpit opened a new issue, #11949:
URL: https://github.com/apache/seatunnel/issues/11949
### Description
The SQL transform functions `lower()` and `upper()` call Java
`String.toLowerCase()` and `String.toUpperCase()` without an explicit locale.
As a result, their output depends on the default locale of the SeaTunnel worker
JVM.
For example, under the Turkish locale, `lower("I")` returns the dotless
character `ı` and `upper("i")` returns `İ`, instead of the locale-independent
SQL results `i` and `I`. A distributed job can therefore produce different
transformed data depending on the host locale.
### Steps to reproduce
1. Check out SeaTunnel `dev`.
2. In `seatunnel-transforms-v2`, invoke the existing `StringFunction`
methods with the JVM default locale set to Turkish:
```java
Locale original = Locale.getDefault();
try {
Locale.setDefault(Locale.forLanguageTag("tr-TR"));
System.out.println(StringFunction.lower(Collections.singletonList("I")));
System.out.println(StringFunction.upper(Collections.singletonList("i")));
} finally {
Locale.setDefault(original);
}
```
3. Run the corresponding test or a small Java harness.
Before the fix, the output is:
```text
lower=ı upper=İ
```
The same harness passes with the proposed fix and produces:
```text
lower=i upper=I
```
### Expected behavior
SQL case conversion should be deterministic and independent of the worker
machine’s default locale. `lower("I")` should return `i` and `upper("i")`
should return `I` regardless of the JVM locale.
### Environment
- SeaTunnel branch: `dev`
- Module: `seatunnel-transforms-v2`
- JVM: Java 8+
- Reproduction does not require an external connector, database, or cluster.
I have prepared a focused fix on the fork branch
`tagadearpit:fix/sql-string-case-locale`, replacing the implicit default locale
with `Locale.ROOT` and adding a regression test. The focused
`StringFunctionTest` passes with 34 tests, and the transforms module
verification lifecycle passes.
--
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]
