[release-18.0] use aggregation engine over distinct engine when overl…

…apping order by (#14359) (#14361) Signed-off-by: Harshit Gangal <[email protected]> Co-authored-by: vitess-bot[bot] <108069721+vitess-bot[bot]@users.noreply.github.com>
vitessio · Oct 25, 2023 · 152b01f · 152b01f
1 parent 40bea23
commit 152b01f
Show file tree

Hide file tree

Showing 3 changed files with 159 additions and 22 deletions.
diff --git a/go/vt/vtgate/planbuilder/operators/queryprojection.go b/go/vt/vtgate/planbuilder/operators/queryprojection.go
@@ -388,8 +388,13 @@ func (qp *QueryProjection) addOrderBy(ctx *plancontext.PlanningContext, orderBy
 
 func (qp *QueryProjection) calculateDistinct(ctx *plancontext.PlanningContext) {
 	if qp.Distinct && !qp.HasAggr {
-		// grouping and distinct both lead to unique results, so we don't need
-		qp.groupByExprs = nil
+		if qp.useGroupingOverDistinct(ctx) {
+			// if order by exists with overlap with select expressions, we can use the aggregation with ordering over distinct.
+			qp.Distinct = false
+		} else {
+			// grouping and distinct both lead to unique results, so we don't need
+			qp.groupByExprs = nil
+		}
 	}
 
 	if qp.HasAggr && len(qp.groupByExprs) == 0 {
@@ -851,6 +856,45 @@ func (qp *QueryProjection) GetColumnCount() int {
 	return len(qp.SelectExprs) - qp.AddedColumn
 }
 
+func (qp *QueryProjection) orderByOverlapWithSelectExpr(ctx *plancontext.PlanningContext) bool {
+	for _, expr := range qp.OrderExprs {
+		idx, _ := qp.FindSelectExprIndexForExpr(ctx, expr.SimplifiedExpr)
+		if idx != nil {
+			return true
+		}
+	}
+	return false
+}
+
+func (qp *QueryProjection) useGroupingOverDistinct(ctx *plancontext.PlanningContext) bool {
+	if !qp.orderByOverlapWithSelectExpr(ctx) {
+		return false
+	}
+	var gbs []GroupBy
+	for idx, selExpr := range qp.SelectExprs {
+		ae, err := selExpr.GetAliasedExpr()
+		if err != nil {
+			// not an alias Expr, cannot continue forward.
+			return false
+		}
+		sExpr := qp.GetSimplifiedExpr(ae.Expr)
+		// check if the grouping already exists on that column.
+		found := slices.IndexFunc(qp.groupByExprs, func(gb GroupBy) bool {
+			return ctx.SemTable.EqualsExprWithDeps(gb.SimplifiedExpr, sExpr)
+		})
+		if found != -1 {
+			continue
+		}
+		groupBy := NewGroupBy(ae.Expr, sExpr, ae)
+		selectExprIdx := idx
+		groupBy.InnerIndex = &selectExprIdx
+
+		gbs = append(gbs, groupBy)
+	}
+	qp.groupByExprs = append(qp.groupByExprs, gbs...)
+	return true
+}
+
 func checkForInvalidGroupingExpressions(expr sqlparser.Expr) error {
 	return sqlparser.Walk(func(node sqlparser.SQLNode) (bool, error) {
 		if _, isAggregate := node.(sqlparser.AggrFunc); isAggregate {

diff --git a/go/vt/vtgate/planbuilder/testdata/oltp_cases.json b/go/vt/vtgate/planbuilder/testdata/oltp_cases.json
@@ -106,28 +106,21 @@
       "QueryType": "SELECT",
       "Original": "SELECT DISTINCT c FROM sbtest30 WHERE id BETWEEN 1 AND 10 ORDER BY c",
       "Instructions": {
-        "OperatorType": "Sort",
-        "Variant": "Memory",
-        "OrderBy": "0 ASC COLLATE latin1_swedish_ci",
+        "OperatorType": "Aggregate",
+        "Variant": "Ordered",
+        "GroupBy": "0 COLLATE latin1_swedish_ci",
         "Inputs": [
           {
-            "OperatorType": "Distinct",
-            "Collations": [
-              "0: latin1_swedish_ci"
-            ],
-            "Inputs": [
-              {
-                "OperatorType": "Route",
-                "Variant": "Scatter",
-                "Keyspace": {
-                  "Name": "main",
-                  "Sharded": true
-                },
-                "FieldQuery": "select c from sbtest30 where 1 != 1",
-                "Query": "select distinct c from sbtest30 where id between 1 and 10",
-                "Table": "sbtest30"
-              }
-            ]
+            "OperatorType": "Route",
+            "Variant": "Scatter",
+            "Keyspace": {
+              "Name": "main",
+              "Sharded": true
+            },
+            "FieldQuery": "select c from sbtest30 where 1 != 1 group by c",
+            "OrderBy": "0 ASC COLLATE latin1_swedish_ci",
+            "Query": "select c from sbtest30 where id between 1 and 10 group by c order by c asc",
+            "Table": "sbtest30"
           }
         ]
       },

diff --git a/go/vt/vtgate/planbuilder/testdata/postprocess_cases.json b/go/vt/vtgate/planbuilder/testdata/postprocess_cases.json
@@ -2081,5 +2081,105 @@
         "user.user"
       ]
     }
+  },
+  {
+    "comment": "distinct with order by using aggregation engine",
+    "query": "select distinct col from user where id between :vtg1 and :vtg2 order by col asc",
+    "plan": {
+      "QueryType": "SELECT",
+      "Original": "select distinct col from user where id between :vtg1 and :vtg2 order by col asc",
+      "Instructions": {
+        "OperatorType": "Aggregate",
+        "Variant": "Ordered",
+        "GroupBy": "0",
+        "Inputs": [
+          {
+            "OperatorType": "Route",
+            "Variant": "Scatter",
+            "Keyspace": {
+              "Name": "user",
+              "Sharded": true
+            },
+            "FieldQuery": "select col from `user` where 1 != 1 group by col",
+            "OrderBy": "0 ASC",
+            "Query": "select col from `user` where id between :vtg1 and :vtg2 group by col order by col asc",
+            "Table": "`user`"
+          }
+        ]
+      },
+      "TablesUsed": [
+        "user.user"
+      ]
+    }
+  },
+  {
+    "comment": "distinct with order by having additional non-order by columns in the selection using aggregation engine",
+    "query": "select distinct foo, col from user where id between :vtg1 and :vtg2 order by col asc",
+    "plan": {
+      "QueryType": "SELECT",
+      "Original": "select distinct foo, col from user where id between :vtg1 and :vtg2 order by col asc",
+      "Instructions": {
+        "OperatorType": "Aggregate",
+        "Variant": "Ordered",
+        "GroupBy": "1, (0|2)",
+        "ResultColumns": 2,
+        "Inputs": [
+          {
+            "OperatorType": "Route",
+            "Variant": "Scatter",
+            "Keyspace": {
+              "Name": "user",
+              "Sharded": true
+            },
+            "FieldQuery": "select foo, col, weight_string(foo) from `user` where 1 != 1 group by col, foo, weight_string(foo)",
+            "OrderBy": "1 ASC, (0|2) ASC",
+            "Query": "select foo, col, weight_string(foo) from `user` where id between :vtg1 and :vtg2 group by col, foo, weight_string(foo) order by col asc, foo asc",
+            "Table": "`user`"
+          }
+        ]
+      },
+      "TablesUsed": [
+        "user.user"
+      ]
+    }
+  },
+  {
+    "comment": "distinct with order by having no overalap with the selection columns - using distinct engine",
+    "query": "select distinct foo from user where id between :vtg1 and :vtg2 order by col asc",
+    "plan": {
+      "QueryType": "SELECT",
+      "Original": "select distinct foo from user where id between :vtg1 and :vtg2 order by col asc",
+      "Instructions": {
+        "OperatorType": "Sort",
+        "Variant": "Memory",
+        "OrderBy": "1 ASC",
+        "ResultColumns": 1,
+        "Inputs": [
+          {
+            "OperatorType": "Distinct",
+            "Collations": [
+              "(0:2)",
+              "1"
+            ],
+            "Inputs": [
+              {
+                "OperatorType": "Route",
+                "Variant": "Scatter",
+                "Keyspace": {
+                  "Name": "user",
+                  "Sharded": true
+                },
+                "FieldQuery": "select foo, col, weight_string(foo) from `user` where 1 != 1",
+                "Query": "select distinct foo, col, weight_string(foo) from `user` where id between :vtg1 and :vtg2",
+                "Table": "`user`"
+              }
+            ]
+          }
+        ]
+      },
+      "TablesUsed": [
+        "user.user"
+      ]
+    }
   }
 ]