The filter of partition including cast function may lead the partition pruning to disable

SQL：select * from test_db.test_table where day between date_sub('2023-12-01',1) and  '2023-12-03'

The Physical Plan of sql above will implement cast function on partition col 'day',  like this, cast(day as date) > 2023-11-30. In this situation, spark just pass the filter condition day < "2023-12-03" to HiveMetastore, not including filter condition cast(day as date) > 2023-11-30, which may lead performance of HMS degarde if the HiveTable has huge number of partitions.

In this regard, a new rule may solve this problem. This rule can convert binary comparison cast(day as date) > 2023-11-30 to day > cast(2023-11-30 as string). The right node is foldable, so the result is day > "2023-11-30", and the filter condition passed to HMS will be day > "2023-11-30" and day < "2023-12-03".