feat: Support EXISTS subquery (#137)

MazterQyou · waralexrom · web-flow · commit fe23133a8351 · 2024-02-28T14:16:54.000+03:00
Co-authored-by: Alexandr Romanenko &lt;alex.romanenko@cube.dev&gt;
diff --git a/datafusion/core/src/logical_plan/builder.rs b/datafusion/core/src/logical_plan/builder.rs
@@ -47,7 +47,8 @@ use super::{dfschema::ToDFSchema, expr_rewriter::coerce_plan_expr_for_schema, Di
 use super::{exprlist_to_fields, Expr, JoinConstraint, JoinType, LogicalPlan, PlanType};
 use crate::logical_plan::{
     columnize_expr, normalize_col, normalize_cols, rewrite_sort_cols_by_aggs, Column,
-    CrossJoin, DFField, DFSchema, DFSchemaRef, Limit, Partitioning, Repartition, Values,
+    CrossJoin, DFField, DFSchema, DFSchemaRef, Limit, Partitioning, Repartition,
+    SubqueryType, Values,
 };
 use crate::sql::utils::group_window_expr_by_sort_keys;
 
@@ -528,12 +529,15 @@ impl LogicalPlanBuilder {
     pub fn subquery(
         &self,
         subqueries: impl IntoIterator<Item = impl Into<LogicalPlan>>,
+        types: impl IntoIterator<Item = SubqueryType>,
     ) -> Result<Self> {
         let subqueries = subqueries.into_iter().map(|l| l.into()).collect::<Vec<_>>();
-        let schema = Arc::new(Subquery::merged_schema(&self.plan, &subqueries));
+        let types = types.into_iter().collect::<Vec<_>>();
+        let schema = Arc::new(Subquery::merged_schema(&self.plan, &subqueries, &types));
         Ok(Self::from(LogicalPlan::Subquery(Subquery {
             input: Arc::new(self.plan.clone()),
             subqueries,
+            types,
             schema,
         })))
     }
diff --git a/datafusion/core/src/logical_plan/mod.rs b/datafusion/core/src/logical_plan/mod.rs
@@ -68,6 +68,6 @@ pub use plan::{
     CreateCatalogSchema, CreateExternalTable, CreateMemoryTable, CrossJoin, Distinct,
     DropTable, EmptyRelation, Filter, JoinConstraint, JoinType, Limit, LogicalPlan,
     Partitioning, PlanType, PlanVisitor, Repartition, StringifiedPlan, Subquery,
-    TableScan, ToStringifiedPlan, Union, Values,
+    SubqueryType, TableScan, ToStringifiedPlan, Union, Values,
 };
 pub use registry::FunctionRegistry;
diff --git a/datafusion/core/src/logical_plan/plan.rs b/datafusion/core/src/logical_plan/plan.rs
@@ -26,7 +26,7 @@ use crate::error::DataFusionError;
 use crate::logical_plan::dfschema::DFSchemaRef;
 use crate::sql::parser::FileType;
 use arrow::datatypes::{DataType, Field, Schema, SchemaRef};
-use datafusion_common::DFSchema;
+use datafusion_common::{DFField, DFSchema};
 use std::fmt::Formatter;
 use std::{
     collections::HashSet,
@@ -267,22 +267,83 @@ pub struct Limit {
 /// Evaluates correlated sub queries
 #[derive(Clone)]
 pub struct Subquery {
-    /// The list of sub queries
-    pub subqueries: Vec<LogicalPlan>,
     /// The incoming logical plan
     pub input: Arc<LogicalPlan>,
+    /// The list of sub queries
+    pub subqueries: Vec<LogicalPlan>,
+    /// The list of subquery types
+    pub types: Vec<SubqueryType>,
     /// The schema description of the output
     pub schema: DFSchemaRef,
 }
 
+/// Subquery type
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub enum SubqueryType {
+    /// Scalar (SELECT, WHERE) evaluating to one value
+    Scalar,
+    /// EXISTS(...) evaluating to true if at least one row was produced
+    Exists,
+    // This will be extended with `AnyAll` type.
+}
+
+impl Display for SubqueryType {
+    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
+        let subquery_type = match self {
+            SubqueryType::Scalar => "Scalar",
+            SubqueryType::Exists => "Exists",
+        };
+        write!(f, "{}", subquery_type)
+    }
+}
+
 impl Subquery {
     /// Merge schema of main input and correlated subquery columns
-    pub fn merged_schema(input: &LogicalPlan, subqueries: &[LogicalPlan]) -> DFSchema {
-        subqueries.iter().fold((**input.schema()).clone(), |a, b| {
-            let mut res = a;
-            res.merge(b.schema());
-            res
-        })
+    pub fn merged_schema(
+        input: &LogicalPlan,
+        subqueries: &[LogicalPlan],
+        types: &[SubqueryType],
+    ) -> DFSchema {
+        subqueries.iter().zip(types.iter()).fold(
+            (**input.schema()).clone(),
+            |schema, (plan, typ)| {
+                let mut schema = schema;
+                schema.merge(&Self::transform_dfschema(plan.schema(), *typ));
+                schema
+            },
+        )
+    }
+
+    /// Transform DataFusion schema according to subquery type
+    pub fn transform_dfschema(schema: &DFSchema, typ: SubqueryType) -> DFSchema {
+        match typ {
+            SubqueryType::Scalar => schema.clone(),
+            SubqueryType::Exists => {
+                let new_fields = schema
+                    .fields()
+                    .iter()
+                    .map(|field| {
+                        let new_field = Subquery::transform_field(field.field(), typ);
+                        if let Some(qualifier) = field.qualifier() {
+                            DFField::from_qualified(qualifier, new_field)
+                        } else {
+                            DFField::from(new_field)
+                        }
+                    })
+                    .collect();
+                DFSchema::new_with_metadata(new_fields, schema.metadata().clone())
+                    .unwrap()
+            } // Schema will be transformed for `AnyAll` as well
+        }
+    }
+
+    /// Transform Arrow field according to subquery type
+    pub fn transform_field(field: &Field, typ: SubqueryType) -> Field {
+        match typ {
+            SubqueryType::Scalar => field.clone(),
+            SubqueryType::Exists => Field::new(field.name(), DataType::Boolean, false),
+            // Field will be transformed for `AnyAll` as well
+        }
     }
 }
 
@@ -475,13 +536,23 @@ impl LogicalPlan {
             LogicalPlan::Values(Values { schema, .. }) => vec![schema],
             LogicalPlan::Window(Window { input, schema, .. })
             | LogicalPlan::Projection(Projection { input, schema, .. })
-            | LogicalPlan::Subquery(Subquery { input, schema, .. })
             | LogicalPlan::Aggregate(Aggregate { input, schema, .. })
             | LogicalPlan::TableUDFs(TableUDFs { input, schema, .. }) => {
                 let mut schemas = input.all_schemas();
                 schemas.insert(0, schema);
                 schemas
             }
+            LogicalPlan::Subquery(Subquery {
+                input,
+                subqueries,
+                schema,
+                ..
+            }) => {
+                let mut schemas = input.all_schemas();
+                schemas.extend(subqueries.iter().map(|s| s.schema()));
+                schemas.insert(0, schema);
+                schemas
+            }
             LogicalPlan::Join(Join {
                 left,
                 right,
@@ -1063,7 +1134,9 @@ impl LogicalPlan {
                         }
                         Ok(())
                     }
-                    LogicalPlan::Subquery(Subquery { .. }) => write!(f, "Subquery"),
+                    LogicalPlan::Subquery(Subquery { types, .. }) => {
+                        write!(f, "Subquery: types={:?}", types)
+                    }
                     LogicalPlan::Filter(Filter {
                         predicate: ref expr,
                         ..
diff --git a/datafusion/core/src/optimizer/projection_drop_out.rs b/datafusion/core/src/optimizer/projection_drop_out.rs
@@ -254,6 +254,7 @@ fn optimize_plan(
         LogicalPlan::Subquery(Subquery {
             input,
             subqueries,
+            types,
             schema,
         }) => {
             // TODO: subqueries are not optimized
@@ -269,6 +270,7 @@ fn optimize_plan(
                         .map(|(p, _)| p)?,
                     ),
                     subqueries: subqueries.clone(),
+                    types: types.clone(),
                     schema: schema.clone(),
                 }),
                 None,
diff --git a/datafusion/core/src/optimizer/projection_push_down.rs b/datafusion/core/src/optimizer/projection_push_down.rs
@@ -453,7 +453,10 @@ fn optimize_plan(
             }))
         }
         LogicalPlan::Subquery(Subquery {
-            input, subqueries, ..
+            input,
+            subqueries,
+            types,
+            ..
         }) => {
             let mut subquery_required_columns = HashSet::new();
             for subquery in subqueries.iter() {
@@ -484,11 +487,12 @@ fn optimize_plan(
                 has_projection,
                 _optimizer_config,
             )?;
-            let new_schema = Subquery::merged_schema(&input, subqueries);
+            let new_schema = Subquery::merged_schema(&input, subqueries, types);
             Ok(LogicalPlan::Subquery(Subquery {
                 input: Arc::new(input),
                 schema: Arc::new(new_schema),
                 subqueries: subqueries.clone(),
+                types: types.clone(),
             }))
         }
         // all other nodes: Add any additional columns used by
diff --git a/datafusion/core/src/optimizer/utils.rs b/datafusion/core/src/optimizer/utils.rs
@@ -161,10 +161,11 @@ pub fn from_plan(
                 alias: alias.clone(),
             }))
         }
-        LogicalPlan::Subquery(Subquery { schema, .. }) => {
+        LogicalPlan::Subquery(Subquery { schema, types, .. }) => {
             Ok(LogicalPlan::Subquery(Subquery {
-                subqueries: inputs[1..inputs.len()].to_vec(),
                 input: Arc::new(inputs[0].clone()),
+                subqueries: inputs[1..inputs.len()].to_vec(),
+                types: types.clone(),
                 schema: schema.clone(),
             }))
         }
diff --git a/datafusion/core/src/physical_plan/planner.rs b/datafusion/core/src/physical_plan/planner.rs
@@ -917,7 +917,7 @@ impl DefaultPhysicalPlanner {
 
                     Ok(Arc::new(GlobalLimitExec::new(input, *skip, *fetch)))
                 }
-                LogicalPlan::Subquery(Subquery { subqueries, input, schema }) => {
+                LogicalPlan::Subquery(Subquery { input, subqueries, types, schema }) => {
                     let cursor = Arc::new(OuterQueryCursor::new(schema.as_ref().to_owned().into()));
                     let mut new_session_state = session_state.clone();
                     new_session_state.execution_props = new_session_state.execution_props.with_outer_query_cursor(cursor.clone());
@@ -931,7 +931,7 @@ impl DefaultPhysicalPlanner {
                         })
                         .collect::<Vec<_>>();
                     let input = self.create_initial_plan(input, &new_session_state).await?;
-                    Ok(Arc::new(SubqueryExec::try_new(subqueries, input, cursor)?))
+                    Ok(Arc::new(SubqueryExec::try_new(input, subqueries, types.clone(), cursor)?))
                 }
                 LogicalPlan::CreateExternalTable(_) => {
                     // There is no default plan for "CREATE EXTERNAL
@@ -1033,6 +1033,7 @@ pub fn create_physical_expr(
             let cursors = execution_props.outer_query_cursors.clone();
             let cursor = cursors
                 .iter()
+                .rev()
                 .find(|cur| cur.schema().field_with_name(c.name.as_str()).is_ok())
                 .ok_or_else(|| {
                     DataFusionError::Execution(format!(
diff --git a/datafusion/core/src/physical_plan/subquery.rs b/datafusion/core/src/physical_plan/subquery.rs
diff --git a/datafusion/core/src/sql/planner.rs b/datafusion/core/src/sql/planner.rs
diff --git a/datafusion/core/tests/sql/subquery.rs b/datafusion/core/tests/sql/subquery.rs

Original file line number	Diff line number	Diff line change
`@@ -161,10 +161,11 @@ pub fn from_plan(`
`161`	`161`	`alias: alias.clone(),`
`162`	`162`	`}))`
`163`	`163`	`}`
`164`		`- LogicalPlan::Subquery(Subquery { schema, .. }) => {`
	`164`	`+ LogicalPlan::Subquery(Subquery { schema, types, .. }) => {`
`165`	`165`	`Ok(LogicalPlan::Subquery(Subquery {`
`166`		`- subqueries: inputs[1..inputs.len()].to_vec(),`
`167`	`166`	`input: Arc::new(inputs[0].clone()),`
	`167`	`+ subqueries: inputs[1..inputs.len()].to_vec(),`
	`168`	`+ types: types.clone(),`
`168`	`169`	`schema: schema.clone(),`
`169`	`170`	`}))`
`170`	`171`	`}`