From 97d5ac34f0f7ef8ee5434fc1d6d89e12deb28a6d Mon Sep 17 00:00:00 2001
From: Fedomn <fedomn.ma@gmail.com>
Date: Tue, 20 Sep 2022 21:43:25 +0800
Subject: [PATCH] feat(distinct): convert select distinct to groupby with no
 aggregations

Signed-off-by: Fedomn <fedomn.ma@gmail.com>
---
 src/binder/statement/mod.rs |  3 +++
 src/planner/mod.rs          | 30 ++++++++++++++++++++++++++++++
 src/planner/select.rs       |  5 +++++
 tests/slt/select.slt        | 15 +++++++++++++++
 4 files changed, 53 insertions(+)

diff --git a/src/binder/statement/mod.rs b/src/binder/statement/mod.rs
index 284661e..2e44836 100644
--- a/src/binder/statement/mod.rs
+++ b/src/binder/statement/mod.rs
@@ -19,6 +19,7 @@ pub struct BoundSelect {
     pub limit: Option<BoundExpr>,
     pub offset: Option<BoundExpr>,
     pub order_by: Vec<BoundOrderBy>,
+    pub select_distinct: bool,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -61,6 +62,7 @@ impl Binder {
                 }
             }
         }
+        let select_distinct = select.distinct;
 
         // bind where clause
         let where_clause = select
@@ -109,6 +111,7 @@ impl Binder {
             limit,
             offset,
             order_by,
+            select_distinct,
         })
     }
 
diff --git a/src/planner/mod.rs b/src/planner/mod.rs
index 93fa563..3ab0b8c 100644
--- a/src/planner/mod.rs
+++ b/src/planner/mod.rs
@@ -30,6 +30,22 @@ mod planner_test {
     };
     use crate::optimizer::PlanNodeType;
 
+    fn build_test_select_distinct_stmt() -> BoundStatement {
+        let c1 = build_bound_column_ref("t", "c1");
+        let t = build_table_ref("t", vec!["c1", "c2"]);
+
+        BoundStatement::Select(BoundSelect {
+            select_list: vec![c1],
+            from_table: Some(t),
+            where_clause: None,
+            group_by: vec![],
+            limit: None,
+            offset: None,
+            order_by: vec![],
+            select_distinct: true,
+        })
+    }
+
     fn build_test_select_stmt() -> BoundStatement {
         let c1 = build_bound_column_ref("t", "c1");
         let t = build_table_ref("t", vec!["c1", "c2"]);
@@ -49,6 +65,7 @@ mod planner_test {
             limit: Some(BoundExpr::Constant(10.into())),
             offset: None,
             order_by: vec![],
+            select_distinct: false,
         })
     }
 
@@ -83,6 +100,7 @@ mod planner_test {
             limit: None,
             offset: None,
             order_by: vec![],
+            select_distinct: false,
         })
     }
 
@@ -152,4 +170,16 @@ mod planner_test {
 
         dbg!(plan_ref);
     }
+
+    #[test]
+    fn test_plan_select_distinct_works() {
+        let stmt = build_test_select_distinct_stmt();
+        let p = Planner {};
+        let node = p.plan(stmt);
+        assert!(node.is_ok());
+        let plan_ref = node.unwrap();
+        assert_eq!(plan_ref.node_type(), PlanNodeType::LogicalProject);
+        assert_eq!(plan_ref.children()[0].node_type(), PlanNodeType::LogicalAgg);
+        dbg!(plan_ref);
+    }
 }
diff --git a/src/planner/select.rs b/src/planner/select.rs
index 66a1cf1..0d84657 100644
--- a/src/planner/select.rs
+++ b/src/planner/select.rs
@@ -26,6 +26,11 @@ impl Planner {
             plan = Arc::new(LogicalAgg::new(agg, stmt.group_by, plan));
         }
 
+        if stmt.select_distinct {
+            // convert distinct to groupby with no aggregations
+            plan = Arc::new(LogicalAgg::new(vec![], stmt.select_list.clone(), plan));
+        }
+
         // LogicalOrder should be below LogicalProject in tree due to it could contains column_ref
         if !stmt.order_by.is_empty() {
             plan = Arc::new(LogicalOrder::new(stmt.order_by, plan));
diff --git a/tests/slt/select.slt b/tests/slt/select.slt
index c946a3c..3d28ead 100644
--- a/tests/slt/select.slt
+++ b/tests/slt/select.slt
@@ -5,3 +5,18 @@ Bill CA 1 12000
 Gregg CO 2 10000
 John CO 3 11500
 Von (empty) 4 NULL
+
+query I
+select distinct state from employee
+----
+CA
+CO
+(empty)
+
+query II
+select distinct a, b from t2
+----
+10 2
+20 2
+30 3
+40 4