Spyros Pavlatos created SPARK-57747:
---------------------------------------
Summary: Fix translate() to distinguish deletion from literal
U+0000 replacement
Key: SPARK-57747
URL: https://issues.apache.org/jira/browse/SPARK-57747
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 4.2.0
Reporter: Spyros Pavlatos
{{translate}} currently treats {{U+0000}} ({{{}\u0000{}}}, {{{}X'00'{}}}) as a
deletion marker in its internal mapping logic. Because of this, a valid request
to replace a character with a literal NUL is misinterpreted as “delete this
character,” producing length 0 instead of length 1 output for that position.
This ticket fixes that specific ambiguity so literal {{U+0000}} replacement is
preserved as data, not interpreted as delete. The change should be applied
consistently across Catalyst dictionary construction,
{{{}UTF8String.translate{}}}, and any codegen/collation translate paths, with
regression tests that assert {{translate(..., ..., '\u0000')}} emits a real NUL
character.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
