planner: correct estimated row count for inner plan of index join (#10015) (#10044)

Author: eurekaka

cmd/explaintest/r/explain_easy.result

@@ -60,8 +60,8 @@ explain select count(b.c2) from t1 a, t2 b where a.c1 = b.c2 group by a.c1;
 id	count	task	operator info
 Projection_11	10000.00	root	cast(join_agg_0)
 └─IndexJoin_14	10000.00	root	inner join, inner:TableReader_13, outer key:b.c2, inner key:a.c1
-  ├─TableReader_13	10.00	root	data:TableScan_12
-  │ └─TableScan_12	10.00	cop	table:a, range: decided by [b.c2], keep order:false, stats:pseudo
+  ├─TableReader_13	1.00	root	data:TableScan_12
+  │ └─TableScan_12	1.00	cop	table:a, range: decided by [b.c2], keep order:false, stats:pseudo
   └─HashAgg_21	8000.00	root	group by:col_2, funcs:count(col_0), firstrow(col_1)
     └─TableReader_22	8000.00	root	data:HashAgg_17
       └─HashAgg_17	8000.00	cop	group by:b.c2, funcs:count(b.c2), firstrow(b.c2)
@@ -279,9 +279,9 @@ Projection_11	10000.00	root	9_aux_0
     └─IndexJoin_44	10000.00	root	inner join, inner:TableReader_43, outer key:s.a, inner key:t1.a
       ├─TableReader_37	1.00	root	data:TableScan_36
       │ └─TableScan_36	1.00	cop	table:s, range: decided by [eq(s.a, test.t.a)], keep order:false, stats:pseudo
-      └─TableReader_43	8000.00	root	data:Selection_42
-        └─Selection_42	8000.00	cop	eq(t1.a, test.t.a)
-          └─TableScan_41	10.00	cop	table:t1, range: decided by [s.a], keep order:false, stats:pseudo
+      └─TableReader_43	0.80	root	data:Selection_42
+        └─Selection_42	0.80	cop	eq(t1.a, test.t.a)
+          └─TableScan_41	1.00	cop	table:t1, range: decided by [s.a], keep order:false, stats:pseudo
 explain select t.c in (select count(*) from t s use index(idx), t t1 where s.b = t.a and s.a = t1.a) from t;
 id	count	task	operator info
 Projection_11	10000.00	root	9_aux_0
@@ -292,8 +292,8 @@ Projection_11	10000.00	root	9_aux_0
     └─IndexJoin_32	10000.00	root	inner join, inner:TableReader_31, outer key:s.a, inner key:t1.a
       ├─IndexReader_27	10000.00	root	index:IndexScan_26
       │ └─IndexScan_26	10000.00	cop	table:s, index:b, range: decided by [eq(s.b, test.t.a)], keep order:false, stats:pseudo
-      └─TableReader_31	10.00	root	data:TableScan_30
-        └─TableScan_30	10.00	cop	table:t1, range: decided by [s.a], keep order:false, stats:pseudo
+      └─TableReader_31	1.00	root	data:TableScan_30
+        └─TableScan_30	1.00	cop	table:t1, range: decided by [s.a], keep order:false, stats:pseudo
 explain select t.c in (select count(*) from t s use index(idx), t t1 where s.b = t.a and s.c = t1.a) from t;
 id	count	task	operator info
 Projection_11	10000.00	root	9_aux_0
@@ -305,8 +305,8 @@ Projection_11	10000.00	root	9_aux_0
       ├─IndexLookUp_28	10000.00	root	
       │ ├─IndexScan_26	10000.00	cop	table:s, index:b, range: decided by [eq(s.b, test.t.a)], keep order:false, stats:pseudo
       │ └─TableScan_27	10000.00	cop	table:t, keep order:false, stats:pseudo
-      └─TableReader_32	10.00	root	data:TableScan_31
-        └─TableScan_31	10.00	cop	table:t1, range: decided by [s.c], keep order:false, stats:pseudo
+      └─TableReader_32	1.00	root	data:TableScan_31
+        └─TableScan_31	1.00	cop	table:t1, range: decided by [s.c], keep order:false, stats:pseudo
 drop table if exists t;
 create table t(a int unsigned);
 explain select t.a = '123455' from t;

cmd/explaintest/r/tpch.result

@@ -601,8 +601,8 @@ Sort_22	2406.00	root	profit.nation:asc, profit.o_year:desc
         │ ├─IndexJoin_40	241379546.70	root	inner join, inner:TableReader_39, outer key:tpch.supplier.s_nationkey, inner key:tpch.nation.n_nationkey
         │ │ ├─IndexJoin_43	241379546.70	root	inner join, inner:TableReader_42, outer key:tpch.lineitem.l_suppkey, inner key:tpch.supplier.s_suppkey
         │ │ │ ├─IndexJoin_47	241379546.70	root	inner join, inner:TableReader_46, outer key:tpch.lineitem.l_partkey, inner key:tpch.part.p_partkey
-        │ │ │ │ ├─TableReader_46	8000000.00	root	data:Selection_45
-        │ │ │ │ │ └─Selection_45	8000000.00	cop	like(tpch.part.p_name, "%dim%", 92)
+        │ │ │ │ ├─TableReader_46	0.80	root	data:Selection_45
+        │ │ │ │ │ └─Selection_45	0.80	cop	like(tpch.part.p_name, "%dim%", 92)
         │ │ │ │ │   └─TableScan_44	1.00	cop	table:part, range: decided by [tpch.lineitem.l_partkey], keep order:false
         │ │ │ │ └─IndexLookUp_51	300005811.00	root	
         │ │ │ │   ├─IndexScan_49	300005811.00	cop	table:lineitem, index:L_ORDERKEY, L_LINENUMBER, range:[NULL,+inf], keep order:true
@@ -1098,8 +1098,8 @@ StreamAgg_13	1.00	root	funcs:sum(col_0)
     ├─TableReader_22	6286493.79	root	data:Selection_21
     │ └─Selection_21	6286493.79	cop	eq(tpch.lineitem.l_shipinstruct, "DELIVER IN PERSON"), in(tpch.lineitem.l_shipmode, "AIR", "AIR REG"), or(and(ge(tpch.lineitem.l_quantity, 4), le(tpch.lineitem.l_quantity, 14)), or(and(ge(tpch.lineitem.l_quantity, 18), le(tpch.lineitem.l_quantity, 28)), and(ge(tpch.lineitem.l_quantity, 29), le(tpch.lineitem.l_quantity, 39))))
     │   └─TableScan_20	300005811.00	cop	table:lineitem, range:[-inf,+inf], keep order:false
-    └─TableReader_28	8000000.00	root	data:Selection_27
-      └─Selection_27	8000000.00	cop	ge(tpch.part.p_size, 1), or(and(eq(tpch.part.p_brand, "Brand#52"), and(in(tpch.part.p_container, "SM CASE", "SM BOX", "SM PACK", "SM PKG"), le(tpch.part.p_size, 5))), or(and(eq(tpch.part.p_brand, "Brand#11"), and(in(tpch.part.p_container, "MED BAG", "MED BOX", "MED PKG", "MED PACK"), le(tpch.part.p_size, 10))), and(eq(tpch.part.p_brand, "Brand#51"), and(in(tpch.part.p_container, "LG CASE", "LG BOX", "LG PACK", "LG PKG"), le(tpch.part.p_size, 15)))))
+    └─TableReader_28	0.80	root	data:Selection_27
+      └─Selection_27	0.80	cop	ge(tpch.part.p_size, 1), or(and(eq(tpch.part.p_brand, "Brand#52"), and(in(tpch.part.p_container, "SM CASE", "SM BOX", "SM PACK", "SM PKG"), le(tpch.part.p_size, 5))), or(and(eq(tpch.part.p_brand, "Brand#11"), and(in(tpch.part.p_container, "MED BAG", "MED BOX", "MED PKG", "MED PACK"), le(tpch.part.p_size, 10))), and(eq(tpch.part.p_brand, "Brand#51"), and(in(tpch.part.p_container, "LG CASE", "LG BOX", "LG PACK", "LG PKG"), le(tpch.part.p_size, 15)))))
         └─TableScan_26	1.00	cop	table:part, range: decided by [tpch.lineitem.l_partkey], keep order:false
 /*
 Q20 Potential Part Promotion Query

planner/core/exhaust_physical_plans.go

@@ -490,22 +490,15 @@ func (p *LogicalJoin) constructInnerTableScan(ds *DataSource, pk *expression.Col
 	}.init(ds.ctx)
 	ts.SetSchema(ds.schema)
 
-	var rowCount float64
-	pkHist, ok := ds.statisticTable.Columns[pk.ID]
-	if ok && !ds.statisticTable.Pseudo {
-		rowCount = pkHist.AvgCountPerValue(ds.statisticTable.Count)
-	} else {
-		rowCount = ds.statisticTable.PseudoAvgCountPerValue()
-	}
-
-	ts.stats = property.NewSimpleStats(rowCount)
+	ts.stats = property.NewSimpleStats(1)
 	ts.stats.UsePseudoStats = ds.statisticTable.Pseudo
 
 	copTask := &copTask{
 		tablePlan:         ts,
 		indexPlanFinished: true,
 	}
-	ts.addPushedDownSelection(copTask, ds.stats)
+	selStats := ts.stats.Scale(selectionFactor)
+	ts.addPushedDownSelection(copTask, selStats)
 	t := finishCopTask(ds.ctx, copTask)
 	return t.plan()
 }