planner: make index join's code about build tableScan and indexScan more clean. (#60107)

ref #60106

Author: AilinKid

pkg/planner/core/exhaust_physical_plans.go

@@ -892,13 +892,13 @@ func buildIndexJoinInner2TableScan(
 			return nil
 		}
 		rangeInfo := indexJoinPathRangeInfo(p.SCtx(), outerJoinKeys, indexJoinResult)
-		innerTask = constructInnerTableScanTask(p, prop, wrapper, indexJoinResult.chosenRanges.Range(), outerJoinKeys, rangeInfo, false, false, avgInnerRowCnt)
+		innerTask = constructInnerTableScanTask(p, prop, wrapper, indexJoinResult.chosenRanges.Range(), rangeInfo, false, false, avgInnerRowCnt)
 		// The index merge join's inner plan is different from index join, so we
 		// should construct another inner plan for it.
 		// Because we can't keep order for union scan, if there is a union scan in inner task,
 		// we can't construct index merge join.
 		if !wrapper.hasDitryWrite {
-			innerTask2 = constructInnerTableScanTask(p, prop, wrapper, indexJoinResult.chosenRanges.Range(), outerJoinKeys, rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt)
+			innerTask2 = constructInnerTableScanTask(p, prop, wrapper, indexJoinResult.chosenRanges.Range(), rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt)
 		}
 		ranges = indexJoinResult.chosenRanges
 	} else {
@@ -932,13 +932,13 @@ func buildIndexJoinInner2TableScan(
 		}
 		buffer.WriteString("]")
 		rangeInfo := buffer.String()
-		innerTask = constructInnerTableScanTask(p, prop, wrapper, ranges, outerJoinKeys, rangeInfo, false, false, avgInnerRowCnt)
+		innerTask = constructInnerTableScanTask(p, prop, wrapper, ranges, rangeInfo, false, false, avgInnerRowCnt)
 		// The index merge join's inner plan is different from index join, so we
 		// should construct another inner plan for it.
 		// Because we can't keep order for union scan, if there is a union scan in inner task,
 		// we can't construct index merge join.
 		if !wrapper.hasDitryWrite {
-			innerTask2 = constructInnerTableScanTask(p, prop, wrapper, ranges, outerJoinKeys, rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt)
+			innerTask2 = constructInnerTableScanTask(p, prop, wrapper, ranges, rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt)
 		}
 	}
 	var (
@@ -1000,7 +1000,7 @@ func buildIndexJoinInner2IndexScan(
 			maxOneRow = ok && (sf.FuncName.L == ast.EQ)
 		}
 	}
-	innerTask := constructInnerIndexScanTask(p, prop, wrapper, indexJoinResult.chosenPath, indexJoinResult.chosenRanges.Range(), indexJoinResult.chosenRemained, innerJoinKeys, indexJoinResult.idxOff2KeyOff, rangeInfo, false, false, avgInnerRowCnt, maxOneRow)
+	innerTask := constructInnerIndexScanTask(p, prop, wrapper, indexJoinResult.chosenPath, indexJoinResult.chosenRanges.Range(), indexJoinResult.chosenRemained, indexJoinResult.idxOff2KeyOff, rangeInfo, false, false, avgInnerRowCnt, maxOneRow)
 	failpoint.Inject("MockOnlyEnableIndexHashJoin", func(val failpoint.Value) {
 		if val.(bool) && !p.SCtx().GetSessionVars().InRestrictedSQL && innerTask != nil {
 			failpoint.Return(constructIndexHashJoin(p, prop, outerIdx, innerTask, indexJoinResult.chosenRanges, keyOff2IdxOff, indexJoinResult.chosenPath, indexJoinResult.lastColManager))
@@ -1017,7 +1017,7 @@ func buildIndexJoinInner2IndexScan(
 	// Because we can't keep order for union scan, if there is a union scan in inner task,
 	// we can't construct index merge join.
 	if !wrapper.hasDitryWrite {
-		innerTask2 := constructInnerIndexScanTask(p, prop, wrapper, indexJoinResult.chosenPath, indexJoinResult.chosenRanges.Range(), indexJoinResult.chosenRemained, innerJoinKeys, indexJoinResult.idxOff2KeyOff, rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt, maxOneRow)
+		innerTask2 := constructInnerIndexScanTask(p, prop, wrapper, indexJoinResult.chosenPath, indexJoinResult.chosenRanges.Range(), indexJoinResult.chosenRemained, indexJoinResult.idxOff2KeyOff, rangeInfo, true, !prop.IsSortItemEmpty() && prop.SortItems[0].Desc, avgInnerRowCnt, maxOneRow)
 		if innerTask2 != nil {
 			joins = append(joins, constructIndexMergeJoin(p, prop, outerIdx, innerTask2, indexJoinResult.chosenRanges, keyOff2IdxOff, indexJoinResult.chosenPath, indexJoinResult.lastColManager)...)
 		}
@@ -1031,13 +1031,27 @@ func constructInnerTableScanTask(
 	prop *property.PhysicalProperty,
 	wrapper *indexJoinInnerChildWrapper,
 	ranges ranger.Ranges,
-	_ []*expression.Column,
 	rangeInfo string,
 	keepOrder bool,
 	desc bool,
 	rowCount float64,
 ) base.Task {
-	ds := wrapper.ds
+	copTask := constructDS2TableScanTask(wrapper.ds, ranges, rangeInfo, keepOrder, desc, rowCount)
+	if copTask == nil {
+		return nil
+	}
+	return constructIndexJoinInnerSideTaskWithAggCheck(p, prop, copTask.(*CopTask), wrapper.ds, nil, wrapper)
+}
+
+// constructInnerTableScanTask is specially used to construct the inner plan for PhysicalIndexJoin.
+func constructDS2TableScanTask(
+	ds *logicalop.DataSource,
+	ranges ranger.Ranges,
+	rangeInfo string,
+	keepOrder bool,
+	desc bool,
+	rowCount float64,
+) base.Task {
 	// If `ds.TableInfo.GetPartitionInfo() != nil`,
 	// it means the data source is a partition table reader.
 	// If the inner task need to keep order, the partition table reader can't satisfy it.
@@ -1082,7 +1096,7 @@ func constructInnerTableScanTask(
 		StatsVersion: ds.StatsInfo().StatsVersion,
 		// NDV would not be used in cost computation of IndexJoin, set leave it as default nil.
 	})
-	usedStats := p.SCtx().GetSessionVars().StmtCtx.GetUsedStatsInfo(false)
+	usedStats := ds.SCtx().GetSessionVars().StmtCtx.GetUsedStatsInfo(false)
 	if usedStats != nil && usedStats.GetUsedInfo(ts.physicalTableID) != nil {
 		ts.usedStatsInfo = usedStats.GetUsedInfo(ts.physicalTableID)
 	}
@@ -1101,7 +1115,7 @@ func constructInnerTableScanTask(
 	ts.PlanPartInfo = copTask.physPlanPartInfo
 	selStats := ts.StatsInfo().Scale(selectivity)
 	ts.addPushedDownSelection(copTask, selStats)
-	return constructIndexJoinInnerSideTaskWithAggCheck(p, prop, copTask, ds, nil, wrapper)
+	return copTask
 }
 
 func constructIndexJoinInnerSideTask(curTask base.Task, prop *property.PhysicalProperty, zippedChildren []base.LogicalPlan, skipAgg bool) base.Task {
@@ -1227,15 +1241,33 @@ func constructInnerIndexScanTask(
 	path *util.AccessPath,
 	ranges ranger.Ranges,
 	filterConds []expression.Expression,
-	_ []*expression.Column,
 	idxOffset2joinKeyOffset []int,
 	rangeInfo string,
 	keepOrder bool,
 	desc bool,
 	rowCount float64,
 	maxOneRow bool,
 ) base.Task {
-	ds := wrapper.ds
+	copTask := constructDS2IndexScanTask(wrapper.ds, path, ranges, filterConds, idxOffset2joinKeyOffset, rangeInfo, keepOrder, desc, rowCount, maxOneRow)
+	if copTask == nil {
+		return nil
+	}
+	return constructIndexJoinInnerSideTaskWithAggCheck(p, prop, copTask.(*CopTask), wrapper.ds, path, wrapper)
+}
+
+// constructDS2IndexScanTask is specially used to construct the inner plan for PhysicalIndexJoin.
+func constructDS2IndexScanTask(
+	ds *logicalop.DataSource,
+	path *util.AccessPath,
+	ranges ranger.Ranges,
+	filterConds []expression.Expression,
+	idxOffset2joinKeyOffset []int,
+	rangeInfo string,
+	keepOrder bool,
+	desc bool,
+	rowCount float64,
+	maxOneRow bool,
+) base.Task {
 	// If `ds.TableInfo.GetPartitionInfo() != nil`,
 	// it means the data source is a partition table reader.
 	// If the inner task need to keep order, the partition table reader can't satisfy it.
@@ -1299,7 +1331,7 @@ func constructInnerIndexScanTask(
 		// change before calling `(*copTask).finishIndexPlan`, we don't know the stats information of `ts` currently and on
 		// the other hand, it may be hard to identify `StatsVersion` of `ts` in `(*copTask).finishIndexPlan`.
 		ts.SetStats(&property.StatsInfo{StatsVersion: ds.TableStats.StatsVersion})
-		usedStats := p.SCtx().GetSessionVars().StmtCtx.GetUsedStatsInfo(false)
+		usedStats := ds.SCtx().GetSessionVars().StmtCtx.GetUsedStatsInfo(false)
 		if usedStats != nil && usedStats.GetUsedInfo(ts.physicalTableID) != nil {
 			ts.usedStatsInfo = usedStats.GetUsedInfo(ts.physicalTableID)
 		}
@@ -1405,7 +1437,7 @@ func constructInnerIndexScanTask(
 		logutil.BgLogger().Warn("unexpected error happened during addPushedDownSelection function", zap.Error(err))
 		return nil
 	}
-	return constructIndexJoinInnerSideTaskWithAggCheck(p, prop, cop, ds, path, wrapper)
+	return cop
 }
 
 // construct the inner join task by inner child plan tree
@@ -1599,6 +1631,7 @@ func tryToGetIndexJoin(p *logicalop.LogicalJoin, prop *property.PhysicalProperty
 		return candidates, canForced
 	}
 	candidates = handleFilterIndexJoinHints(p, candidates)
+	// todo: if any variables banned it, why bother to generate it first?
 	return filterIndexJoinBySessionVars(p.SCtx(), candidates), false
 }