Query: Don't add grouping key to projection when Distinct is applied (#…

…28072)

Resolves #28039

src/EFCore.Relational/Query/SqlExpressions/SelectExpression.cs

@@ -2993,7 +2993,8 @@ private SqlRemappingVisitor PushdownIntoSubqueryInternal()
             }
         }
 
-        if (subquery._groupBy.Count > 0)
+        if (subquery._groupBy.Count > 0
+            && !subquery.IsDistinct)
         {
             foreach (var key in subquery._groupBy)
             {

test/EFCore.Specification.Tests/Query/NorthwindGroupByQueryTestBase.cs

@@ -1854,6 +1854,21 @@ from g in grouping
                 },
             entryCount: 63);
 
+    [ConditionalTheory]
+    [MemberData(nameof(IsAsyncData))]
+    public virtual Task Join_GroupBy_Aggregate_distinct_single_join(bool async)
+        => AssertQuery(
+            async,
+            ss =>
+                from c in ss.Set<Customer>()
+                join a in ss.Set<Order>().GroupBy(o => new { o.CustomerID, o.OrderDate.Value.Year } )
+                        .Where(g => g.Count() > 5)
+                        .Select(g => new { CustomerID = g.Key.CustomerID, LastOrderID = g.Max(o => o.OrderID) })
+                        .Distinct()
+                    on c.CustomerID equals a.CustomerID
+                select new { c, a.LastOrderID },
+            entryCount: 31);
+
     [ConditionalTheory]
     [MemberData(nameof(IsAsyncData))]
     public virtual Task Join_GroupBy_Aggregate_with_left_join(bool async)

test/EFCore.Specification.Tests/Query/SimpleQueryTestBase.cs

@@ -1134,4 +1134,68 @@ protected class Child26744
         public DateTime? SomeOtherNullableDateTime { get; set; }
         public Parent26744 Parent { get; set; }
     }
+
+    [ConditionalTheory]
+    [MemberData(nameof(IsAsyncData))]
+    public virtual async Task Pushdown_does_not_add_grouping_key_to_projection_when_distinct_is_applied(bool async)
+    {
+        var contextFactory = await InitializeAsync<Context28039>();
+        using var db = contextFactory.CreateContext();
+
+
+        var queryResults = (from i in db.IndexData.Where(a => a.Parcel == "some condition")
+                                        .Select(a => new SearchResult { ParcelNumber = a.Parcel, RowId = a.RowId })
+                            group i by new { i.ParcelNumber, i.RowId } into grp
+                            where grp.Count() == 1
+                            select grp.Key.ParcelNumber).Distinct();
+
+        var jsonLookup = (from dcv in db.TableData.Where(a => a.TableId == 123)
+                          join wos in queryResults
+                          on dcv.ParcelNumber equals wos
+                          orderby dcv.ParcelNumber
+                          select dcv.JSON).Take(123456);
+
+
+        var result = async
+            ? await jsonLookup.ToListAsync()
+            : jsonLookup.ToList();
+    }
+
+    protected class Context28039 : DbContext
+    {
+        public Context28039(DbContextOptions options)
+            : base(options)
+        {
+        }
+
+        public DbSet<IndexData> IndexData { get; set; }
+        public DbSet<TableData> TableData { get; set; }
+    }
+
+    public class TableData : EntityBase
+    {
+        public int TableId { get; set; }
+        public string ParcelNumber { get; set; }
+        public short RowId { get; set; }
+        public string JSON { get; set; }
+
+    }
+
+    public abstract class EntityBase
+    {
+        [Key]
+        public int ID { get; set; }
+    }
+    public class IndexData : EntityBase
+    {
+        public string Parcel { get; set; }
+        public int RowId { get; set; }
+    }
+
+    internal class SearchResult
+    {
+        public string ParcelNumber { get; set; }
+        public int RowId { get; set; }
+        public string DistinctValue { get; set; }
+    }
 }

test/EFCore.SqlServer.FunctionalTests/Query/NorthwindGroupByQuerySqlServerTest.cs

@@ -1589,6 +1589,24 @@ HAVING COUNT(*) > 5
 INNER JOIN [Orders] AS [o0] ON [c].[CustomerID] = [o0].[CustomerID]");
     }
 
+    public override async Task Join_GroupBy_Aggregate_distinct_single_join(bool async)
+    {
+        await base.Join_GroupBy_Aggregate_distinct_single_join(async);
+
+        AssertSql(
+            @"SELECT [c].[CustomerID], [c].[Address], [c].[City], [c].[CompanyName], [c].[ContactName], [c].[ContactTitle], [c].[Country], [c].[Fax], [c].[Phone], [c].[PostalCode], [c].[Region], [t0].[LastOrderID]
+FROM [Customers] AS [c]
+INNER JOIN (
+    SELECT DISTINCT [t].[CustomerID], MAX([t].[OrderID]) AS [LastOrderID]
+    FROM (
+        SELECT [o].[OrderID], [o].[CustomerID], DATEPART(year, [o].[OrderDate]) AS [Year]
+        FROM [Orders] AS [o]
+    ) AS [t]
+    GROUP BY [t].[CustomerID], [t].[Year]
+    HAVING COUNT(*) > 5
+) AS [t0] ON [c].[CustomerID] = [t0].[CustomerID]");
+    }
+
     public override async Task Join_GroupBy_Aggregate_with_left_join(bool async)
     {
         await base.Join_GroupBy_Aggregate_with_left_join(async);

test/EFCore.SqlServer.FunctionalTests/Query/SimpleQuerySqlServerTest.cs

@@ -409,4 +409,24 @@ FROM [MyEntities] AS [m]
 FROM [MyEntities] AS [m]
 WHERE [dbo].[ModifyDate]([m].[SomeDate]) = @__date_0");
     }
+
+    public override async Task Pushdown_does_not_add_grouping_key_to_projection_when_distinct_is_applied(bool async)
+    {
+        await base.Pushdown_does_not_add_grouping_key_to_projection_when_distinct_is_applied(async);
+
+        AssertSql(
+            @"@__p_0='123456'
+
+SELECT TOP(@__p_0) [t].[JSON]
+FROM [TableData] AS [t]
+INNER JOIN (
+    SELECT DISTINCT [i].[Parcel]
+    FROM [IndexData] AS [i]
+    WHERE [i].[Parcel] = N'some condition'
+    GROUP BY [i].[Parcel], [i].[RowId]
+    HAVING COUNT(*) = 1
+) AS [t0] ON [t].[ParcelNumber] = [t0].[Parcel]
+WHERE [t].[TableId] = 123
+ORDER BY [t].[ParcelNumber]");
+    }
 }