Linhong Liu created SPARK-60574:
-----------------------------------
Summary: Quote collated string literals in connector expressions
Key: SPARK-60574
URL: https://issues.apache.org/jira/browse/SPARK-60574
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 5.0.0
Reporter: Linhong Liu
The connector LiteralValue SQL renderer matches the StringType singleton rather
than all StringType instances. Non-default collations therefore miss SQL
quoting and single-quote escaping. For example,
LiteralValue(UTF8String.fromString("it's"), StringType("UTF8_LCASE")).toString
renders it's instead of 'it''s'.
STARTS_WITH, ENDS_WITH and CONTAINS descriptions also assume that string
literals have surrounding quotes. An empty collated string can cause a
substring exception, and longer strings can lose their first and last
characters.
Use a StringType type pattern to apply the existing quoting behavior to all
string types. Keep NULL rendering unchanged. Cover empty, short and quoted
strings, LIKE metacharacters, collations and constrained string types.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
