build_join_top

datafusion_optimizer::decorrelate_predicate_subquery

Function build_join_top 

Source 
fn build_join_top(
    query_info: &SubqueryInfo,
    left: &LogicalPlan,
    alias: &Arc<AliasGenerator>,
) -> Result<Option<LogicalPlan>>
Expand description

Optimize the subquery to left-anti/left-semi join. If the subquery is a correlated subquery, we need extract the join predicate from the subquery.

For example, given a query like: select t1.a, t1.b from t1 where t1 in (select t2.a from t2 where t1.b = t2.b and t1.c > t2.c)

The optimized plan will be:

Projection: t1.a, t1.b
  LeftSemi Join:  Filter: t1.a = __correlated_sq_1.a AND t1.b = __correlated_sq_1.b AND t1.c > __correlated_sq_1.c
    TableScan: t1
    SubqueryAlias: __correlated_sq_1
      Projection: t2.a, t2.b, t2.c
        TableScan: t2

Given another query like: select t1.id from t1 where exists(SELECT t2.id FROM t2 WHERE t1.id = t2.id)

The optimized plan will be:

Projection: t1.id
  LeftSemi Join:  Filter: t1.id = __correlated_sq_1.id
    TableScan: t1
    SubqueryAlias: __correlated_sq_1
      Projection: t2.id
        TableScan: t2