Skip to content

Commit

Permalink
compute large and
Browse files Browse the repository at this point in the history
  • Loading branch information
@englefly
englefly committed Nov 11, 2024
1 parent 2e28db9 commit e2305d2
Show file tree
Hide file tree
Showing 68 changed files with 984 additions and 937 deletions.
114 changes: 83 additions & 31 deletions fe/fe-core/src/main/java/org/apache/doris/nereids/stats/FilterEstimation.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -46,6 +46,7 @@
import org.apache.doris.nereids.types.DataType;
import org.apache.doris.nereids.types.DateTimeType;
import org.apache.doris.nereids.types.coercion.RangeScalable;
import org.apache.doris.nereids.util.ExpressionUtils;
import org.apache.doris.statistics.ColumnStatistic;
import org.apache.doris.statistics.ColumnStatisticBuilder;
import org.apache.doris.statistics.StatisticRange;
Expand All @@ -60,6 +61,7 @@
import java.util.Map;
import java.util.Optional;
import java.util.Set;
import java.util.stream.Collectors;

/**
* Calculate selectivity of expression that produces boolean value.
Expand Down Expand Up @@ -112,42 +114,89 @@ public Statistics visit(Expression expr, EstimationContext context) {
return context.statistics.withSel(DEFAULT_INEQUALITY_COEFFICIENT);
}

private Statistics estimateLargeAnd(List<Expression> conjuncts, EstimationContext context) {
StatisticsBuilder builder = new StatisticsBuilder(context.statistics);
if (context.statistics.getRowCount() <= 1) {
return builder.build();
}
List<Double> orderedSelections = conjuncts.stream()
.mapToDouble(conjunct ->
conjunct.accept(this, context).getBENumber() / context.statistics.getRowCount())
.boxed()
.sorted()
.collect(Collectors.toList());

double sel = 1.0;
for (int i = 0; i < orderedSelections.size(); i++) {
double value = orderedSelections.get(i);
if (value >= 1) {
break;
}
double root = Math.pow(value, 1.0 / (i + 1));
sel *= root;
}
double inputRowCount = context.statistics.getRowCount();
Statistics outputStats = builder.setRowCount(inputRowCount * sel).build();
return outputStats.normalizeColumnStatistics(inputRowCount, true);
}

private Statistics estimateBasicAnd(List<Expression> conjuncts, EstimationContext context) {
Expression leftExpr = conjuncts.get(0);
Expression rightExpr = conjuncts.get(1);
Statistics leftStats = leftExpr.accept(this, context);
leftStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
Statistics andStats = rightExpr.accept(this, new EstimationContext(leftStats));
return andStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
}

@Override
public Statistics visitCompoundPredicate(CompoundPredicate predicate, EstimationContext context) {
Expression leftExpr = predicate.child(0);
Expression rightExpr = predicate.child(1);
public Statistics visitAnd(And and, EstimationContext context) {
List<Expression> conjuncts = ExpressionUtils.extractConjunction(and);
Statistics outputStats;
if (conjuncts.size() >= LARGE_COMPOUND_PREDICATE) {
outputStats = estimateLargeAnd(conjuncts, context);
} else {
outputStats = estimateBasicAnd(conjuncts, context);
}
return outputStats;
}

@Override
public Statistics visitOr(Or or, EstimationContext context) {
Expression leftExpr = or.child(0);
Expression rightExpr = or.child(1);
Statistics leftStats = leftExpr.accept(this, context);
leftStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
Statistics andStats = rightExpr.accept(this, new EstimationContext(leftStats));
if (predicate instanceof And) {
andStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
return andStats;
} else if (predicate instanceof Or) {
Statistics rightStats = rightExpr.accept(this, context);
rightStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
double rowCount = leftStats.getRowCount() + rightStats.getRowCount() - andStats.getRowCount();
Statistics orStats = context.statistics.withRowCount(rowCount);
Set<Slot> leftInputSlots = leftExpr.getInputSlots();
Set<Slot> rightInputSlots = rightExpr.getInputSlots();
for (Slot slot : context.keyColumns) {
if (leftInputSlots.contains(slot) && rightInputSlots.contains(slot)) {
ColumnStatistic leftColStats = leftStats.findColumnStatistics(slot);
ColumnStatistic rightColStats = rightStats.findColumnStatistics(slot);
StatisticRange leftRange = StatisticRange.from(leftColStats, slot.getDataType());
StatisticRange rightRange = StatisticRange.from(rightColStats, slot.getDataType());
StatisticRange union = leftRange.union(rightRange);
ColumnStatisticBuilder colBuilder = new ColumnStatisticBuilder(
context.statistics.findColumnStatistics(slot));
colBuilder.setMinValue(union.getLow()).setMinExpr(union.getLowExpr())
.setMaxValue(union.getHigh()).setMaxExpr(union.getHighExpr())
.setNdv(union.getDistinctValues());
double maxNumNulls = Math.max(leftColStats.numNulls, rightColStats.numNulls);
colBuilder.setNumNulls(Math.min(colBuilder.getCount(), maxNumNulls));
orStats.addColumnStats(slot, colBuilder.build());
}

Statistics rightStats = rightExpr.accept(this, context);
rightStats.normalizeColumnStatistics(context.statistics.getRowCount(), true);
double rowCount = leftStats.getRowCount() + rightStats.getRowCount() - andStats.getRowCount();
Statistics orStats = context.statistics.withRowCount(rowCount);
Set<Slot> leftInputSlots = leftExpr.getInputSlots();
Set<Slot> rightInputSlots = rightExpr.getInputSlots();
for (Slot slot : context.keyColumns) {
if (leftInputSlots.contains(slot) && rightInputSlots.contains(slot)) {
ColumnStatistic leftColStats = leftStats.findColumnStatistics(slot);
ColumnStatistic rightColStats = rightStats.findColumnStatistics(slot);
StatisticRange leftRange = StatisticRange.from(leftColStats, slot.getDataType());
StatisticRange rightRange = StatisticRange.from(rightColStats, slot.getDataType());
StatisticRange union = leftRange.union(rightRange);
ColumnStatisticBuilder colBuilder = new ColumnStatisticBuilder(
context.statistics.findColumnStatistics(slot));
colBuilder.setMinValue(union.getLow()).setMinExpr(union.getLowExpr())
.setMaxValue(union.getHigh()).setMaxExpr(union.getHighExpr())
.setNdv(union.getDistinctValues());
double maxNumNulls = Math.max(leftColStats.numNulls, rightColStats.numNulls);
colBuilder.setNumNulls(Math.min(colBuilder.getCount(), maxNumNulls));
orStats.addColumnStats(slot, colBuilder.build());
}
return orStats;
}
return orStats;
}

@Override
public Statistics visitCompoundPredicate(CompoundPredicate predicate, EstimationContext context) {
// should not come here
Preconditions.checkArgument(false,
"unsupported compound operator: %s in %s",
Expand Down Expand Up @@ -577,7 +626,10 @@ public Statistics visitIsNull(IsNull isNull, EstimationContext context) {
return builder.build();
}

static class EstimationContext {
/**
* EstimationContext
*/
public static class EstimationContext {
private final Statistics statistics;

private final Set<Slot> keyColumns = Sets.newHashSet();
Expand Down
3 changes: 2 additions & 1 deletion fe/fe-core/src/main/java/org/apache/doris/statistics/Statistics.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -112,7 +112,7 @@ public void normalizeColumnStatistics(double inputRowCount) {
normalizeColumnStatistics(this.rowCount, false);
}

public void normalizeColumnStatistics(double inputRowCount, boolean isNumNullsDecreaseByProportion) {
public Statistics normalizeColumnStatistics(double inputRowCount, boolean isNumNullsDecreaseByProportion) {
double factor = isNumNullsDecreaseByProportion ? rowCount / inputRowCount : 1.0;
for (Entry<Expression, ColumnStatistic> entry : expressionToColumnStats.entrySet()) {
ColumnStatistic columnStatistic = entry.getValue();
Expand All @@ -130,6 +130,7 @@ public void normalizeColumnStatistics(double inputRowCount, boolean isNumNullsDe
expressionToColumnStats.put(entry.getKey(), columnStatistic);
}
}
return this;
}

public boolean checkColumnStatsValid(ColumnStatistic columnStatistic, double rowCount) {
Expand Down
20 changes: 10 additions & 10 deletions regression-test/data/nereids_tpcds_shape_sf100_p0/noStatsRfPrune/query11.out
View file
Original file line number Diff line number Diff line change
Expand Up @@ -35,20 +35,20 @@ PhysicalCteAnchor ( cteId=CTEId#0 )
------PhysicalDistribute[DistributionSpecGather]
--------PhysicalTopN[LOCAL_SORT]
----------PhysicalProject
------------hashJoin[INNER_JOIN shuffleBucket] hashCondition=((t_s_firstyear.customer_id = t_w_secyear.customer_id)) otherCondition=((if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000) > if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000))) build RFs:RF5 customer_id->[customer_id]
------------hashJoin[INNER_JOIN broadcast] hashCondition=((t_s_firstyear.customer_id = t_w_firstyear.customer_id)) otherCondition=((if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000) > if((year_total > 0.00), (cast(year_total as DECIMALV3(38, 8)) / year_total), 0.000000))) build RFs:RF5 customer_id->[customer_id,customer_id,customer_id]
--------------PhysicalProject
----------------filter((t_w_secyear.dyear = 2002) and (t_w_secyear.sale_type = 'w'))
------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF5
--------------PhysicalProject
----------------hashJoin[INNER_JOIN bucketShuffle] hashCondition=((t_s_firstyear.customer_id = t_w_firstyear.customer_id)) otherCondition=() build RFs:RF4 customer_id->[customer_id,customer_id]
----------------hashJoin[INNER_JOIN shuffleBucket] hashCondition=((t_s_firstyear.customer_id = t_w_secyear.customer_id)) otherCondition=() build RFs:RF4 customer_id->[customer_id]
------------------PhysicalProject
--------------------filter((t_w_secyear.dyear = 2002) and (t_w_secyear.sale_type = 'w'))
----------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF4 RF5
------------------hashJoin[INNER_JOIN shuffle] hashCondition=((t_s_secyear.customer_id = t_s_firstyear.customer_id)) otherCondition=() build RFs:RF3 customer_id->[customer_id]
--------------------PhysicalProject
----------------------filter((t_s_secyear.dyear = 2002) and (t_s_secyear.sale_type = 's'))
------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF3 RF4
------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF3 RF5
--------------------PhysicalProject
----------------------filter((t_s_firstyear.dyear = 2001) and (t_s_firstyear.sale_type = 's') and (t_s_firstyear.year_total > 0.00))
------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF4
------------------PhysicalProject
--------------------filter((t_w_firstyear.dyear = 2001) and (t_w_firstyear.sale_type = 'w') and (t_w_firstyear.year_total > 0.00))
----------------------PhysicalCteConsumer ( cteId=CTEId#0 )
------------------------PhysicalCteConsumer ( cteId=CTEId#0 ) apply RFs: RF5
--------------PhysicalProject
----------------filter((t_w_firstyear.dyear = 2001) and (t_w_firstyear.sale_type = 'w') and (t_w_firstyear.year_total > 0.00))
------------------PhysicalCteConsumer ( cteId=CTEId#0 )

24 changes: 12 additions & 12 deletions regression-test/data/nereids_tpcds_shape_sf100_p0/noStatsRfPrune/query13.out
View file
Original file line number Diff line number Diff line change
Expand Up @@ -7,27 +7,27 @@ PhysicalResultSink
--------PhysicalProject
----------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_sold_date_sk = date_dim.d_date_sk)) otherCondition=() build RFs:RF4 d_date_sk->[ss_sold_date_sk]
------------PhysicalProject
--------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_addr_sk = customer_address.ca_address_sk)) otherCondition=((((ca_state IN ('KS', 'MI', 'SD') AND ((store_sales.ss_net_profit >= 100.00) AND (store_sales.ss_net_profit <= 200.00))) OR (ca_state IN ('CO', 'MO', 'ND') AND ((store_sales.ss_net_profit >= 150.00) AND (store_sales.ss_net_profit <= 300.00)))) OR (ca_state IN ('NH', 'OH', 'TX') AND ((store_sales.ss_net_profit >= 50.00) AND (store_sales.ss_net_profit <= 250.00))))) build RFs:RF3 ca_address_sk->[ss_addr_sk]
--------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_addr_sk = customer_address.ca_address_sk)) otherCondition=((((ca_state IN ('KS', 'MI', 'SD') AND ((store_sales.ss_net_profit >= 100.00) AND (store_sales.ss_net_profit <= 200.00))) OR (ca_state IN ('CO', 'MO', 'ND') AND ((store_sales.ss_net_profit >= 150.00) AND (store_sales.ss_net_profit <= 300.00)))) OR (ca_state IN ('NH', 'OH', 'TX') AND ((store_sales.ss_net_profit >= 50.00) AND (store_sales.ss_net_profit <= 250.00))))) build RFs:RF3 ss_addr_sk->[ca_address_sk]
----------------PhysicalProject
------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_hdemo_sk = household_demographics.hd_demo_sk)) otherCondition=((((((household_demographics.hd_dep_count = 1) AND cd_marital_status IN ('M', 'S')) AND cd_education_status IN ('4 yr Degree', 'College')) AND ((((customer_demographics.cd_marital_status = 'S') AND (customer_demographics.cd_education_status = 'College')) AND ((store_sales.ss_sales_price >= 50.00) AND (store_sales.ss_sales_price <= 100.00))) OR (((customer_demographics.cd_marital_status = 'M') AND (customer_demographics.cd_education_status = '4 yr Degree')) AND ((store_sales.ss_sales_price >= 150.00) AND (store_sales.ss_sales_price <= 200.00))))) OR ((((customer_demographics.cd_marital_status = 'D') AND (customer_demographics.cd_education_status = 'Unknown')) AND ((store_sales.ss_sales_price >= 100.00) AND (store_sales.ss_sales_price <= 150.00))) AND (household_demographics.hd_dep_count = 3)))) build RFs:RF2 hd_demo_sk->[ss_hdemo_sk]
------------------filter((customer_address.ca_country = 'United States') and ca_state IN ('CO', 'KS', 'MI', 'MO', 'ND', 'NH', 'OH', 'SD', 'TX'))
--------------------PhysicalOlapScan[customer_address] apply RFs: RF3
----------------PhysicalProject
------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store_sales.ss_hdemo_sk = household_demographics.hd_demo_sk)) otherCondition=((((((household_demographics.hd_dep_count = 1) AND cd_marital_status IN ('M', 'S')) AND cd_education_status IN ('4 yr Degree', 'College')) AND ((((customer_demographics.cd_marital_status = 'S') AND (customer_demographics.cd_education_status = 'College')) AND ((store_sales.ss_sales_price >= 50.00) AND (store_sales.ss_sales_price <= 100.00))) OR (((customer_demographics.cd_marital_status = 'M') AND (customer_demographics.cd_education_status = '4 yr Degree')) AND ((store_sales.ss_sales_price >= 150.00) AND (store_sales.ss_sales_price <= 200.00))))) OR ((((customer_demographics.cd_marital_status = 'D') AND (customer_demographics.cd_education_status = 'Unknown')) AND ((store_sales.ss_sales_price >= 100.00) AND (store_sales.ss_sales_price <= 150.00))) AND (household_demographics.hd_dep_count = 3)))) build RFs:RF2 ss_hdemo_sk->[hd_demo_sk]
--------------------PhysicalProject
----------------------filter(hd_dep_count IN (1, 3))
------------------------PhysicalOlapScan[household_demographics] apply RFs: RF2
--------------------PhysicalProject
----------------------hashJoin[INNER_JOIN broadcast] hashCondition=((customer_demographics.cd_demo_sk = store_sales.ss_cdemo_sk)) otherCondition=() build RFs:RF1 cd_demo_sk->[ss_cdemo_sk]
------------------------PhysicalProject
--------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store.s_store_sk = store_sales.ss_store_sk)) otherCondition=()
--------------------------hashJoin[INNER_JOIN broadcast] hashCondition=((store.s_store_sk = store_sales.ss_store_sk)) otherCondition=() build RFs:RF0 ss_store_sk->[s_store_sk]
----------------------------PhysicalProject
------------------------------filter((store_sales.ss_net_profit <= 300.00) and (store_sales.ss_net_profit >= 50.00) and (store_sales.ss_sales_price <= 200.00) and (store_sales.ss_sales_price >= 50.00))
--------------------------------PhysicalOlapScan[store_sales] apply RFs: RF1 RF2 RF3 RF4
------------------------------PhysicalOlapScan[store] apply RFs: RF0
----------------------------PhysicalProject
------------------------------PhysicalOlapScan[store]
------------------------------filter((store_sales.ss_net_profit <= 300.00) and (store_sales.ss_net_profit >= 50.00) and (store_sales.ss_sales_price <= 200.00) and (store_sales.ss_sales_price >= 50.00))
--------------------------------PhysicalOlapScan[store_sales] apply RFs: RF1 RF4
------------------------PhysicalProject
--------------------------filter(((((customer_demographics.cd_marital_status = 'D') AND (customer_demographics.cd_education_status = 'Unknown')) OR ((customer_demographics.cd_marital_status = 'S') AND (customer_demographics.cd_education_status = 'College'))) OR ((customer_demographics.cd_marital_status = 'M') AND (customer_demographics.cd_education_status = '4 yr Degree'))) and cd_education_status IN ('4 yr Degree', 'College', 'Unknown') and cd_marital_status IN ('D', 'M', 'S'))
----------------------------PhysicalOlapScan[customer_demographics]
--------------------PhysicalProject
----------------------filter(hd_dep_count IN (1, 3))
------------------------PhysicalOlapScan[household_demographics]
----------------PhysicalProject
------------------filter((customer_address.ca_country = 'United States') and ca_state IN ('CO', 'KS', 'MI', 'MO', 'ND', 'NH', 'OH', 'SD', 'TX'))
--------------------PhysicalOlapScan[customer_address]
------------PhysicalProject
--------------filter((date_dim.d_year = 2001))
----------------PhysicalOlapScan[date_dim]
Expand Down
Loading

0 comments on commit e2305d2

Please sign in to comment.