org.apache.hadoop.hive.ql.udf.ptf

Class BasePartitionEvaluator

java.lang.Object

org.apache.hadoop.hive.ql.udf.ptf.BasePartitionEvaluator

Direct Known Subclasses:

BasePartitionEvaluator.AvgPartitionEvaluator, BasePartitionEvaluator.SumPartitionEvaluator

public class BasePartitionEvaluator
extends Object

This class is mostly used for RANGE windowing type to do some optimization. ROWS windowing type can support streaming.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	BasePartitionEvaluator.AvgPartitionDoubleEvaluator
static class	BasePartitionEvaluator.AvgPartitionEvaluator<ResultType extends org.apache.hadoop.io.Writable> The partition evalulator for average function
static class	BasePartitionEvaluator.AvgPartitionHiveDecimalEvaluator
static class	BasePartitionEvaluator.SumPartitionDoubleEvaluator
static class	BasePartitionEvaluator.SumPartitionEvaluator<ResultType extends org.apache.hadoop.io.Writable> The base type for sum operator evaluator when a partition data is available and streaming process is not possible.
static class	BasePartitionEvaluator.SumPartitionHiveDecimalEvaluator
static class	BasePartitionEvaluator.SumPartitionLongEvaluator

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	isCountEvaluator
protected static org.slf4j.Logger	LOG
protected boolean	nullsLast
protected ObjectInspector	outputOI
protected List<PTFExpressionDef>	parameters
protected PTFPartition	partition
protected WindowFrameDef	winFrame
protected GenericUDAFEvaluator	wrappedEvaluator

Constructor Summary

Constructors

Constructor and Description
BasePartitionEvaluator(GenericUDAFEvaluator wrappedEvaluator, WindowFrameDef winFrame, PTFPartition partition, List<PTFExpressionDef> parameters, ObjectInspector outputOI, boolean nullsLast)

Method Summary

Modifier and Type	Method and Description
protected Object	calcFunctionValue(PTFPartition.PTFPartitionIterator<Object> pItr, LeadLagInfo leadLagInfo) Given a partition iterator, calculate the function value
Object	getPartitionAgg() Get the aggregation for the whole partition.
Object	iterate(int currentRow, LeadLagInfo leadLagInfo) Given the current row, get the aggregation for the window
protected Range	newRange(int end, int end2, PTFPartition partition) When there are no parameters specified, partition iterator can be made faster; In such cases, it need not materialize the ROW from RowContainer.

Methods inherited from class java.lang.Object

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

Field Detail

LOG

protected static final org.slf4j.Logger LOG

wrappedEvaluator

protected final GenericUDAFEvaluator wrappedEvaluator

winFrame

protected final WindowFrameDef winFrame

partition

protected final PTFPartition partition

parameters

protected final List<PTFExpressionDef> parameters

outputOI

protected final ObjectInspector outputOI

nullsLast

protected final boolean nullsLast

isCountEvaluator

protected final boolean isCountEvaluator

Constructor Detail

BasePartitionEvaluator

public BasePartitionEvaluator(GenericUDAFEvaluator wrappedEvaluator,
                              WindowFrameDef winFrame,
                              PTFPartition partition,
                              List<PTFExpressionDef> parameters,
                              ObjectInspector outputOI,
                              boolean nullsLast)

Method Detail

getPartitionAgg

public Object getPartitionAgg()
                       throws HiveException

Get the aggregation for the whole partition. Used in the case where windowing is unbounded or the function value is calculated based on all the rows in the partition such as percent_rank().

Returns:

the aggregated result

Throws:

HiveException

iterate

public Object iterate(int currentRow,
                      LeadLagInfo leadLagInfo)
               throws HiveException

Given the current row, get the aggregation for the window

Throws:

HiveException

calcFunctionValue

protected Object calcFunctionValue(PTFPartition.PTFPartitionIterator<Object> pItr,
                                   LeadLagInfo leadLagInfo)
                            throws HiveException

Given a partition iterator, calculate the function value

Parameters:

pItr - the partition pointer

Returns:

the function value

Throws:

HiveException

newRange

protected Range newRange(int end,
                         int end2,
                         PTFPartition partition)

When there are no parameters specified, partition iterator can be made faster; In such cases, it need not materialize the ROW from RowContainer. This saves lots of IO.