org.apache.drill.exec.expr.stat

Class ParquetPredicates.EqualPredicate

java.lang.Object

org.apache.drill.common.expression.LogicalExpressionBase

org.apache.drill.exec.expr.stat.ParquetPredicates.ParquetCompPredicate

org.apache.drill.exec.expr.stat.ParquetPredicates.EqualPredicate

All Implemented Interfaces:

Iterable<LogicalExpression>, LogicalExpression, ParquetFilterPredicate

Enclosing class:

ParquetPredicates

public static class ParquetPredicates.EqualPredicate
extends ParquetPredicates.ParquetCompPredicate

EQ (=) predicate

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.drill.common.expression.LogicalExpression

LogicalExpression.De, LogicalExpression.Se

Field Summary

Fields inherited from class org.apache.drill.exec.expr.stat.ParquetPredicates.ParquetCompPredicate

left, right

Fields inherited from interface org.apache.drill.common.expression.LogicalExpression

logger

Constructor Summary

Constructors

Constructor and Description
ParquetPredicates.EqualPredicate(LogicalExpression left, LogicalExpression right)

Method Summary

Methods

Modifier and Type	Method and Description
boolean	canDrop(RangeExprEvaluator evaluator) Semantics of canDrop() is very similar to what is implemented in Parquet library's org.apache.parquet.filter2.statisticslevel.StatisticsFilter and FilterPredicate Main difference : 1.
String	toString()

Methods inherited from class org.apache.drill.exec.expr.stat.ParquetPredicates.ParquetCompPredicate

accept, iterator

Methods inherited from class org.apache.drill.common.expression.LogicalExpressionBase

getCumulativeCost, getDescription, getMajorType, getPosition, getSelfCost, i

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

ParquetPredicates.EqualPredicate

public ParquetPredicates.EqualPredicate(LogicalExpression left,
                                LogicalExpression right)

Method Detail

canDrop

public boolean canDrop(RangeExprEvaluator evaluator)

Semantics of canDrop() is very similar to what is implemented in Parquet library's org.apache.parquet.filter2.statisticslevel.StatisticsFilter and FilterPredicate Main difference : 1. A RangeExprEvaluator is used to compute the min/max of an expression, such as CAST function of a column. CAST function could be explicitly added by Drill user (It's recommended to use CAST function after DRILL-4372, if user wants to reduce planning time for limit 0 query), or implicitly inserted by Drill, when the types of compare operands are not identical. Therefore, it's important to allow CAST function to appear in the filter predicate. 2. We do not require list of ColumnChunkMetaData to do the evaluation, while Parquet library's StatisticsFilter has such requirement. Drill's ParquetTableMetaData does not maintain ColumnChunkMetaData, making it impossible to directly use Parquet library's StatisticFilter in query planning time. 3. We allows both sides of comparison operator to be a min/max range. As such, we support expression_of(Column1) < expression_of(Column2), where Column1 and Column2 are from same parquet table.

toString

public String toString()

Overrides:

toString in class Object