planner: avoid change the type of the input of sum/avg (pingcap#36372)

close pingcap#36371

cmd/explaintest/r/agg_predicate_pushdown.result

@@ -7,7 +7,7 @@ id	estRows	task	access object	operator info
 Projection	711.11	root		test.t.a, test.t.b, Column#5
 └─Selection	711.11	root		gt(Column#5, 3)
   └─HashAgg	888.89	root		group by:Column#16, Column#17, Column#18, funcs:avg(Column#13)->Column#5, funcs:firstrow(Column#14)->test.t.a, funcs:firstrow(Column#15)->test.t.b
-    └─Projection	1111.11	root		cast(test.t.c, decimal(15,4) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
+    └─Projection	1111.11	root		cast(test.t.c, decimal(20,0) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
       └─TableReader	1111.11	root		data:Selection
         └─Selection	1111.11	cop[tikv]		gt(test.t.a, 1), gt(test.t.a, 2), gt(test.t.b, 2)
           └─TableFullScan	10000.00	cop[tikv]	table:t	keep order:false, stats:pseudo
@@ -17,7 +17,7 @@ id	estRows	task	access object	operator info
 Projection	657.65	root		test.t.a, test.t.b, Column#5
 └─Selection	657.65	root		gt(Column#5, 3)
   └─HashAgg	822.06	root		group by:Column#16, Column#17, Column#18, funcs:avg(Column#13)->Column#5, funcs:firstrow(Column#14)->test.t.a, funcs:firstrow(Column#15)->test.t.b
-    └─Projection	1027.57	root		cast(test.t.c, decimal(15,4) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
+    └─Projection	1027.57	root		cast(test.t.c, decimal(20,0) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
       └─TableReader	1027.57	root		data:Selection
         └─Selection	1027.57	cop[tikv]		gt(test.t.b, 2), or(gt(test.t.a, 1), gt(test.t.b, 2)), or(gt(test.t.a, 2), lt(test.t.b, 1))
           └─TableFullScan	10000.00	cop[tikv]	table:t	keep order:false, stats:pseudo
@@ -27,7 +27,7 @@ id	estRows	task	access object	operator info
 Projection	3027.54	root		test.t.a, test.t.b, Column#5
 └─Selection	3027.54	root		or(and(gt(test.t.a, 1), gt(test.t.b, 2)), or(and(gt(test.t.a, 2), lt(test.t.b, 1)), and(gt(test.t.b, 2), gt(Column#5, 3))))
   └─HashAgg	3784.43	root		group by:Column#16, Column#17, Column#18, funcs:avg(Column#13)->Column#5, funcs:firstrow(Column#14)->test.t.a, funcs:firstrow(Column#15)->test.t.b
-    └─Projection	4730.53	root		cast(test.t.c, decimal(15,4) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
+    └─Projection	4730.53	root		cast(test.t.c, decimal(20,0) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
       └─TableReader	4730.53	root		data:Selection
         └─Selection	4730.53	cop[tikv]		or(and(gt(test.t.a, 1), gt(test.t.b, 2)), or(and(gt(test.t.a, 2), lt(test.t.b, 1)), gt(test.t.b, 2)))
           └─TableFullScan	10000.00	cop[tikv]	table:t	keep order:false, stats:pseudo
@@ -37,7 +37,7 @@ id	estRows	task	access object	operator info
 Projection	2126.93	root		test.t.a, test.t.b, Column#5
 └─Selection	2126.93	root		or(gt(test.t.a, 1), gt(Column#5, 1))
   └─HashAgg	2658.67	root		group by:Column#16, Column#17, Column#18, funcs:avg(Column#13)->Column#5, funcs:firstrow(Column#14)->test.t.a, funcs:firstrow(Column#15)->test.t.b
-    └─Projection	3323.33	root		cast(test.t.c, decimal(15,4) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
+    └─Projection	3323.33	root		cast(test.t.c, decimal(20,0) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
       └─TableReader	3323.33	root		data:Selection
         └─Selection	3323.33	cop[tikv]		lt(test.t.a, 3)
           └─TableFullScan	10000.00	cop[tikv]	table:t	keep order:false, stats:pseudo
@@ -47,7 +47,7 @@ id	estRows	task	access object	operator info
 Projection	6393.60	root		test.t.a, test.t.b, Column#5
 └─Selection	6393.60	root		or(and(gt(test.t.a, 1), gt(Column#5, 1)), lt(test.t.a, 3))
   └─HashAgg	7992.00	root		group by:Column#16, Column#17, Column#18, funcs:avg(Column#13)->Column#5, funcs:firstrow(Column#14)->test.t.a, funcs:firstrow(Column#15)->test.t.b
-    └─Projection	9990.00	root		cast(test.t.c, decimal(15,4) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
+    └─Projection	9990.00	root		cast(test.t.c, decimal(20,0) BINARY)->Column#13, test.t.a, test.t.b, test.t.a, test.t.b, test.t.c
       └─TableReader	9990.00	root		data:Selection
         └─Selection	9990.00	cop[tikv]		or(gt(test.t.a, 1), lt(test.t.a, 3))
           └─TableFullScan	10000.00	cop[tikv]	table:t	keep order:false, stats:pseudo

cmd/explaintest/r/explain_easy.result

@@ -763,7 +763,7 @@ explain format = 'brief' select a, b from (select a, b, avg(b) over (partition b
 id	estRows	task	access object	operator info
 Projection	2666.67	root		test.t.a, test.t.b
 └─Selection	2666.67	root		gt(cast(test.t.a, decimal(20,0) BINARY), Column#5), lt(test.t.b, 10)
-  └─Window	3333.33	root		avg(cast(test.t.b, decimal(15,4) BINARY))->Column#5 over(partition by test.t.a)
+  └─Window	3333.33	root		avg(cast(test.t.b, decimal(20,0) BINARY))->Column#5 over(partition by test.t.a)
     └─Sort	3333.33	root		test.t.a
       └─TableReader	3333.33	root		data:Selection
         └─Selection	3333.33	cop[tikv]		gt(test.t.a, 10)
@@ -925,7 +925,7 @@ StreamAgg	1.00	root		funcs:sum(Column#14)->Column#13
 explain format = 'brief' select avg(t1.a) from t t1 join t t2 on t1.a=t2.a;
 id	estRows	task	access object	operator info
 StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
-└─Projection	12487.50	root		cast(test.t.a, decimal(8,4) BINARY)->Column#14
+└─Projection	12487.50	root		cast(test.t.a, decimal(20,0) BINARY)->Column#14
   └─HashJoin	12487.50	root		inner join, equal:[eq(test.t.a, test.t.a)]
     ├─TableReader(Build)	9990.00	root		data:Selection
     │ └─Selection	9990.00	cop[tikv]		not(isnull(test.t.a))
@@ -936,7 +936,7 @@ StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
 explain format = 'brief' select avg(t1.b) from t t1 join t t2 on t1.b=t2.b;
 id	estRows	task	access object	operator info
 StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
-└─Projection	12487.50	root		cast(test.t.b, decimal(10,4) BINARY)->Column#14
+└─Projection	12487.50	root		cast(test.t.b, decimal(20,0) BINARY)->Column#14
   └─HashJoin	12487.50	root		inner join, equal:[eq(test.t.b, test.t.b)]
     ├─TableReader(Build)	9990.00	root		data:Selection
     │ └─Selection	9990.00	cop[tikv]		not(isnull(test.t.b))
@@ -947,7 +947,7 @@ StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
 explain format = 'brief' select avg(t1.c) from t t1 join t t2 on t1.c=t2.c;
 id	estRows	task	access object	operator info
 StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
-└─Projection	12487.50	root		cast(test.t.c, decimal(13,4) BINARY)->Column#14
+└─Projection	12487.50	root		cast(test.t.c, decimal(20,0) BINARY)->Column#14
   └─HashJoin	12487.50	root		inner join, equal:[eq(test.t.c, test.t.c)]
     ├─TableReader(Build)	9990.00	root		data:Selection
     │ └─Selection	9990.00	cop[tikv]		not(isnull(test.t.c))
@@ -958,7 +958,7 @@ StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
 explain format = 'brief' select avg(t1.d) from t t1 join t t2 on t1.d=t2.d;
 id	estRows	task	access object	operator info
 StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
-└─Projection	12487.50	root		cast(test.t.d, decimal(15,4) BINARY)->Column#14
+└─Projection	12487.50	root		cast(test.t.d, decimal(20,0) BINARY)->Column#14
   └─HashJoin	12487.50	root		inner join, equal:[eq(test.t.d, test.t.d)]
     ├─TableReader(Build)	9990.00	root		data:Selection
     │ └─Selection	9990.00	cop[tikv]		not(isnull(test.t.d))
@@ -969,7 +969,7 @@ StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
 explain format = 'brief' select avg(t1.e) from t t1 join t t2 on t1.e=t2.e;
 id	estRows	task	access object	operator info
 StreamAgg	1.00	root		funcs:avg(Column#14)->Column#13
-└─Projection	12487.50	root		cast(test.t.e, decimal(24,4) BINARY)->Column#14
+└─Projection	12487.50	root		cast(test.t.e, decimal(20,0) BINARY)->Column#14
   └─HashJoin	12487.50	root		inner join, equal:[eq(test.t.e, test.t.e)]
     ├─TableReader(Build)	9990.00	root		data:Selection
     │ └─Selection	9990.00	cop[tikv]		not(isnull(test.t.e))

cmd/explaintest/r/index_merge.result

@@ -536,7 +536,7 @@ Sort_16	1841.86	root		test.t1.c1
     │     └─TableRowIDScan_24	5542.21	cop[tikv]	table:t1	keep order:false, stats:pseudo
     └─TopN_29(Probe)	1.00	root		test.t2.c1, offset:2, count:1
       └─HashAgg_36	2660.44	root		group by:Column#21, funcs:avg(Column#19)->Column#9, funcs:firstrow(Column#20)->test.t2.c1
-        └─Projection_48	3325.55	root		cast(test.t2.c1, decimal(15,4) BINARY)->Column#19, test.t2.c1, test.t2.c1
+        └─Projection_48	3325.55	root		cast(test.t2.c1, decimal(20,0) BINARY)->Column#19, test.t2.c1, test.t2.c1
           └─IndexMerge_41	3325.55	root		
             ├─Selection_38(Build)	3.32	cop[tikv]		eq(test.t1.c1, test.t2.c1)
             │ └─IndexRangeScan_37	3323.33	cop[tikv]	table:t2, index:c1(c1)	range:[-inf,10), keep order:false, stats:pseudo

expression/aggregation/agg_to_pb_test.go

@@ -53,7 +53,7 @@ func TestAggFunc2Pb(t *testing.T) {
 
 	jsons := []string{
 		`{"tp":3002,"children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":%v,"aggFuncMode":0}`,
-		`{"tp":3001,"children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":8,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":%v,"aggFuncMode":0}`,
+		`{"tp":3001,"children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":8,"flag":0,"flen":20,"decimal":0,"collate":-63,"charset":"binary"},"has_distinct":%v,"aggFuncMode":0}`,
 		`{"tp":3003,"children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":%v,"aggFuncMode":0}`,
 		`{"tp":3007,"val":"AAAAAAAABAA=","children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":15,"flag":0,"flen":-1,"decimal":-1,"collate":-46,"charset":"utf8mb4"},"has_distinct":%v,"aggFuncMode":0}`,
 		`{"tp":3005,"children":[{"tp":201,"val":"gAAAAAAAAAE=","sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":false}],"sig":0,"field_type":{"tp":5,"flag":0,"flen":-1,"decimal":-1,"collate":-63,"charset":"binary"},"has_distinct":%v,"aggFuncMode":0}`,

expression/aggregation/base_func.go

@@ -423,28 +423,6 @@ func (a *baseFuncDesc) WrapCastForAggArgs(ctx sessionctx.Context) {
 		}
 		tpOld := a.Args[i].GetType().GetType()
 		a.Args[i] = castFunc(ctx, a.Args[i])
-		if a.Name != ast.AggFuncAvg && a.Name != ast.AggFuncSum {
-			continue
-		}
-		// After wrapping cast on the argument, flen etc. may not the same
-		// as the type of the aggregation function. The following part set
-		// the type of the argument exactly as the type of the aggregation
-		// function.
-		// Note: If the `tp` of argument is the same as the `tp` of the
-		// aggregation function, it will not wrap cast function on it
-		// internally. The reason of the special handling for `Column` is
-		// that the `RetType` of `Column` refers to the `infoschema`, so we
-		// need to set a new variable for it to avoid modifying the
-		// definition in `infoschema`.
-		if col, ok := a.Args[i].(*expression.Column); ok {
-			col.RetType = types.NewFieldType(col.RetType.GetType())
-		}
-		// originTp is used when the `tp` of column is TypeFloat32 while
-		// the type of the aggregation function is TypeFloat64.
-		originTp := a.Args[i].GetType().GetType()
-		*(a.Args[i].GetType()) = *(a.RetTp)
-		a.Args[i].GetType().SetType(originTp)
-
 		// refine each mysql integer type to the needed decimal precision for sum
 		if a.Name == ast.AggFuncSum {
 			adjustDecimalLenForSumInteger(a.Args[i].GetType(), tpOld)

expression/builtin_cast.go

@@ -1941,7 +1941,9 @@ func WrapWithCastAsDecimal(ctx sessionctx.Context, expr Expression) Expression {
 	tp.SetDecimalUnderLimit(expr.GetType().GetDecimal())
 
 	if expr.GetType().EvalType() == types.ETInt {
+		// todo set the flen returned by minimalDecimalLenForHoldingInteger
 		tp.SetFlen(mysql.MaxIntWidth)
+		tp.SetDecimal(0)
 	}
 	if tp.GetFlen() == types.UnspecifiedLength || tp.GetFlen() > mysql.MaxDecimalWidth {
 		tp.SetFlen(mysql.MaxDecimalWidth)