Yu Xu created CALCITE-7433:
------------------------------
Summary: Invalid unparse for cast to map type in Spark
Key: CALCITE-7433
URL: https://issues.apache.org/jira/browse/CALCITE-7433
Project: Calcite
Issue Type: Bug
Components: core
Affects Versions: 1.41.0
Reporter: Yu Xu
Assignee: Yu Xu
Fix For: 1.42.0
>From https://issues.apache.org/jira/browse/CALCITE-7048 indicates that Spark
>dialects require converting VARCHAR to STRING type, but this conversion fails
>when converting Map type. The test results are as follows:
{code:java}
@Test void testCastMapSpark() {
final String query = "select cast(MAP['a',1,'b',2,'c',3]" + " as
MAP<varchar,integer>)";
final String expectedSpark = "SELECT CAST(MAP ('a', 1, 'b', 2, 'c', 3) AS
MAP< STRING, INTEGER >)\n"
+ "FROM (VALUES (0)) `t` (`ZERO`)";
sql(query).withSpark().ok(expectedSpark);
} {code}
would error：
{code:java}
Expected: is "SELECT CAST(MAP ('a', 1, 'b', 2, 'c', 3) AS MAP< STRING, INTEGER
>)\nFROM (VALUES (0)) `t` (`ZERO`)"
but: was "SELECT CAST(MAP ('a', 1, 'b', 2, 'c', 3) AS MAP< VARCHAR, INTEGER
NULL >)\nFROM (VALUES (0)) `t` (`ZERO`)"
{code}
In the conversion of Map type: STRING is still converted to VARCHAR, which is
an incorrect conversion.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
