#include <aggregate-functions.h>

Collaboration diagram for impala::AggregateFunctions:

Public Member Functions
template<>
void	InitZero (FunctionContext , DecimalVal dst)

template<>
void	Min (FunctionContext ctx, const StringVal &src, StringVal dst)

template<>
void	Max (FunctionContext ctx, const StringVal &src, StringVal dst)

template<>
void	Min (FunctionContext ctx, const DecimalVal &src, DecimalVal dst)

template<>
void	Max (FunctionContext ctx, const DecimalVal &src, DecimalVal dst)

template<>
void	Min (FunctionContext , const TimestampVal &src, TimestampVal dst)

template<>
void	Max (FunctionContext , const TimestampVal &src, TimestampVal dst)

template<>
void	LastValUpdate (FunctionContext ctx, const StringVal &src, StringVal dst)

template<>
void	LastValRemove (FunctionContext ctx, const StringVal &src, StringVal dst)

template<>
void	FirstValUpdate (FunctionContext ctx, const StringVal &src, StringVal dst)

Static Public Member Functions
static void	InitNull (FunctionContext , AnyVal dst)
	Initializes dst to NULL. More...

static void	InitNullString (FunctionContext c, StringVal dst)
	Initializes dst to NULL and sets dst->ptr to NULL. More...

template<typename T >
static void	InitZero (FunctionContext , T dst)
	Initializes dst to 0. More...

static StringVal	StringValGetValue (FunctionContext *ctx, const StringVal &src)
	StringVal GetValue() function that returns a copy of src. More...

static StringVal	StringValSerializeOrFinalize (FunctionContext *ctx, const StringVal &src)
	StringVal Serialize/Finalize function that copies and frees src. More...

static void	CountUpdate (FunctionContext , const AnyVal &src, BigIntVal dst)
	Implementation of Count and Count(*) More...

static void	CountStarUpdate (FunctionContext , BigIntVal dst)

static void	CountRemove (FunctionContext , const AnyVal &src, BigIntVal dst)

static void	CountStarRemove (FunctionContext , BigIntVal dst)

static void	CountMerge (FunctionContext , const BigIntVal &src, BigIntVal dst)

static void	AvgInit (FunctionContext ctx, StringVal dst)

template<typename T >
static void	AvgUpdate (FunctionContext ctx, const T &src, StringVal dst)

template<typename T >
static void	AvgRemove (FunctionContext ctx, const T &src, StringVal dst)

static void	AvgMerge (FunctionContext ctx, const StringVal &src, StringVal dst)

static DoubleVal	AvgGetValue (FunctionContext *ctx, const StringVal &val)

static DoubleVal	AvgFinalize (FunctionContext *ctx, const StringVal &val)

static void	TimestampAvgUpdate (FunctionContext ctx, const TimestampVal &src, StringVal dst)
	Avg for timestamp. Uses AvgInit() and AvgMerge(). More...

static void	TimestampAvgRemove (FunctionContext ctx, const TimestampVal &src, StringVal dst)

static TimestampVal	TimestampAvgGetValue (FunctionContext *ctx, const StringVal &val)

static TimestampVal	TimestampAvgFinalize (FunctionContext *ctx, const StringVal &val)

static void	DecimalAvgInit (FunctionContext ctx, StringVal dst)
	Avg for decimals. More...

static void	DecimalAvgUpdate (FunctionContext ctx, const DecimalVal &src, StringVal dst)

static void	DecimalAvgRemove (FunctionContext ctx, const DecimalVal &src, StringVal dst)

static void	DecimalAvgAddOrRemove (FunctionContext ctx, const DecimalVal &src, StringVal dst, bool remove=false)

static void	DecimalAvgMerge (FunctionContext ctx, const StringVal &src, StringVal dst)

static DecimalVal	DecimalAvgGetValue (FunctionContext *ctx, const StringVal &val)

static DecimalVal	DecimalAvgFinalize (FunctionContext *ctx, const StringVal &val)

template<typename SRC_VAL , typename DST_VAL >
static void	SumUpdate (FunctionContext , const SRC_VAL &src, DST_VAL dst)
	SumUpdate, SumMerge. More...

template<typename SRC_VAL , typename DST_VAL >
static void	SumRemove (FunctionContext , const SRC_VAL &src, DST_VAL dst)

static void	SumDecimalUpdate (FunctionContext , const DecimalVal &src, DecimalVal dst)
	Sum for decimals. More...

static void	SumDecimalRemove (FunctionContext , const DecimalVal &src, DecimalVal dst)

static void	SumDecimalMerge (FunctionContext , const DecimalVal &src, DecimalVal dst)

static void	SumDecimalAddOrSubtract (FunctionContext , const DecimalVal &src, DecimalVal dst, bool subtract=false)
	Adds or or subtracts src from dst. Implements Update() and Remove(). More...

template<typename T >
static void	Min (FunctionContext , const T &src, T dst)
	MinUpdate/MinMerge. More...

template<typename T >
static void	Max (FunctionContext , const T &src, T dst)
	MaxUpdate/MaxMerge. More...

static void	StringConcatUpdate (FunctionContext , const StringVal &src, StringVal result)
	String concat. More...

static void	StringConcatUpdate (FunctionContext , const StringVal &src, const StringVal &separator, StringVal result)

static void	StringConcatMerge (FunctionContext , const StringVal &src, StringVal result)

static StringVal	StringConcatFinalize (FunctionContext *, const StringVal &src)

static void	PcInit (FunctionContext , StringVal slot)

template<typename T >
static void	PcUpdate (FunctionContext , const T &src, StringVal dst)

template<typename T >
static void	PcsaUpdate (FunctionContext , const T &src, StringVal dst)

static void	PcMerge (FunctionContext , const StringVal &src, StringVal dst)

static BigIntVal	PcFinalize (FunctionContext *, const StringVal &src)

static BigIntVal	PcsaFinalize (FunctionContext *, const StringVal &src)

template<typename T >
static void	ReservoirSampleInit (FunctionContext , StringVal slot)

template<typename T >
static void	ReservoirSampleUpdate (FunctionContext , const T &src, StringVal dst)

template<typename T >
static void	ReservoirSampleMerge (FunctionContext , const StringVal &src, StringVal dst)

template<typename T >
static const StringVal	ReservoirSampleSerialize (FunctionContext *, const StringVal &src)

template<typename T >
static StringVal	ReservoirSampleFinalize (FunctionContext *, const StringVal &src)
	Returns 20,000 unsorted samples as a list of comma-separated values. More...

template<typename T >
static T	AppxMedianFinalize (FunctionContext *, const StringVal &src)
	Returns an approximate median using reservoir sampling. More...

template<typename T >
static StringVal	HistogramFinalize (FunctionContext *, const StringVal &src)

static void	HllInit (FunctionContext , StringVal slot)

template<typename T >
static void	HllUpdate (FunctionContext , const T &src, StringVal dst)

static void	HllMerge (FunctionContext , const StringVal &src, StringVal dst)

static BigIntVal	HllFinalize (FunctionContext *, const StringVal &src)

static uint64_t	HllFinalEstimate (const uint8_t *buckets, int32_t num_buckets)

static void	KnuthVarInit (FunctionContext context, StringVal val)

template<typename T >
static void	KnuthVarUpdate (FunctionContext context, const T &input, StringVal val)

static void	KnuthVarMerge (FunctionContext context, const StringVal &src, StringVal dst)

static DoubleVal	KnuthVarFinalize (FunctionContext *context, const StringVal &val)

static DoubleVal	KnuthVarPopFinalize (FunctionContext *context, const StringVal &val)
	Calculates the biased variance, uses KnuthVar Init-Update-Merge functions. More...

static DoubleVal	KnuthStddevFinalize (FunctionContext *context, const StringVal &val)
	Calculates STDDEV, uses KnuthVar Init-Update-Merge functions. More...

static DoubleVal	KnuthStddevPopFinalize (FunctionContext *context, const StringVal &val)
	Calculates the biased STDDEV, uses KnuthVar Init-Update-Merge functions. More...

static void	RankInit (FunctionContext , StringVal slot)
	Initializes the state for RANK and DENSE_RANK. More...

static void	RankUpdate (FunctionContext , StringVal dst)
	Update state for RANK. More...

static void	DenseRankUpdate (FunctionContext , StringVal dst)
	Update state for DENSE_RANK. More...

static BigIntVal	RankGetValue (FunctionContext *, StringVal &src)
	Returns the result for RANK and prepares the state for the next Update(). More...

static BigIntVal	DenseRankGetValue (FunctionContext *, StringVal &src)

static BigIntVal	RankFinalize (FunctionContext *, StringVal &src)
	Returns the result for RANK and DENSE_RANK and cleans up intermediate state in src. More...

template<typename T >
static void	LastValUpdate (FunctionContext , const T &src, T dst)
	Implements LAST_VALUE. More...

template<typename T >
static void	LastValRemove (FunctionContext , const T &src, T dst)

template<typename T >
static void	FirstValUpdate (FunctionContext , const T &src, T dst)
	Implements FIRST_VALUE. More...

template<typename T >
static void	FirstValRewriteUpdate (FunctionContext , const T &src, const BigIntVal &, T dst)

template<typename T >
static void	OffsetFnInit (FunctionContext , T dst)

template<typename T >
static void	OffsetFnUpdate (FunctionContext , const T &src, const BigIntVal &, const T &, T dst)

Static Public Attributes
static const int	HLL_PRECISION = 10

static const int	HLL_LEN = 1024

Detailed Description

Collection of builtin aggregate functions. Aggregate functions implement the various phases of the aggregation: Init(), Update(), Serialize(), Merge(), and Finalize(). Not all functions need to implement all of the steps and some of the parts can be reused across different aggregate functions. This functions are implemented using the UDA interface.

Definition at line 30 of file aggregate-functions.h.

Member Function Documentation

template<typename T >

template DecimalVal impala::AggregateFunctions::AppxMedianFinalize< DecimalVal >	(	FunctionContext *	,
		const StringVal &	src
	)

static

Returns an approximate median using reservoir sampling.

Definition at line 981 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, and impala::ReservoirSampleState< T >::samples.

DoubleVal impala::AggregateFunctions::AvgFinalize	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 177 of file aggregate-functions.cc.

References AvgGetValue(), impala_udf::FunctionContext::Free(), and impala_udf::StringVal::ptr.

DoubleVal impala::AggregateFunctions::AvgGetValue	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 171 of file aggregate-functions.cc.

References impala::AvgState::count, impala_udf::StringVal::ptr, and impala::AvgState::sum.

Referenced by AvgFinalize().

void impala::AggregateFunctions::AvgInit	(	FunctionContext *	ctx,
		StringVal *	dst
	)

static

Implementation of Avg. TODO: Change this to use a fixed-sized BufferVal as intermediate type.

Definition at line 131 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::AvgMerge	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 161 of file aggregate-functions.cc.

References impala::AvgState::count, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::AvgState::sum.

template<typename T >

template void impala::AggregateFunctions::AvgRemove< DoubleVal >	(	FunctionContext *	ctx,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 149 of file aggregate-functions.cc.

References impala::AvgState::count, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::AvgState::sum.

template<typename T >

template void impala::AggregateFunctions::AvgUpdate< DoubleVal >	(	FunctionContext *	ctx,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 139 of file aggregate-functions.cc.

References impala::AvgState::count, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::AvgState::sum.

void impala::AggregateFunctions::CountMerge	(	FunctionContext *	,
		const BigIntVal &	src,
		BigIntVal *	dst
	)

static

Definition at line 119 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::BigIntVal::val.

void impala::AggregateFunctions::CountRemove	(	FunctionContext *	,
		const AnyVal &	src,
		BigIntVal *	dst
	)

static

Definition at line 104 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::BigIntVal::val.

void impala::AggregateFunctions::CountStarRemove	(	FunctionContext *	,
		BigIntVal *	dst
	)

static

Definition at line 113 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::BigIntVal::val.

void impala::AggregateFunctions::CountStarUpdate	(	FunctionContext *	,
		BigIntVal *	dst
	)

static

Definition at line 99 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::BigIntVal::val.

void impala::AggregateFunctions::CountUpdate	(	FunctionContext *	,
		const AnyVal &	src,
		BigIntVal *	dst
	)

static

Implementation of Count and Count(*)

Definition at line 93 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::BigIntVal::val.

void impala::AggregateFunctions::DecimalAvgAddOrRemove	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		StringVal *	dst,
		bool	remove = `false`
	)

static

Definition at line 245 of file aggregate-functions.cc.

References impala::DecimalAvgState::count, impala::ColumnType::DebugString(), impala_udf::FunctionContext::GetArgType(), impala::ColumnType::GetByteSize(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, impala::DecimalAvgState::sum, impala::AnyValUtil::TypeDescToColumnType(), impala_udf::DecimalVal::val16, impala_udf::DecimalVal::val4, and impala_udf::DecimalVal::val8.

Referenced by DecimalAvgRemove(), and DecimalAvgUpdate().

DecimalVal impala::AggregateFunctions::DecimalAvgFinalize	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 315 of file aggregate-functions.cc.

References DecimalAvgGetValue(), impala_udf::FunctionContext::Free(), and impala_udf::StringVal::ptr.

DecimalVal impala::AggregateFunctions::DecimalAvgGetValue	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 290 of file aggregate-functions.cc.

References impala_udf::FunctionContext::AddWarning(), count, impala::DecimalAvgState::count, impala::ColumnType::CreateDecimalType(), impala::DecimalValue< T >::Divide(), impala_udf::FunctionContext::GetReturnType(), impala_udf::StringVal::ptr, impala_udf::FunctionContext::TypeDesc::scale, impala::DecimalAvgState::sum, impala_udf::FunctionContext::TypeDesc::type, impala::TYPE_DECIMAL, UNLIKELY, impala_udf::DecimalVal::val16, and impala::DecimalValue< T >::value().

Referenced by DecimalAvgFinalize().

void impala::AggregateFunctions::DecimalAvgInit	(	FunctionContext *	ctx,
		StringVal *	dst
	)

static

Avg for decimals.

Definition at line 228 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::DecimalAvgMerge	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 279 of file aggregate-functions.cc.

References impala::DecimalAvgState::count, impala_udf::StringVal::len, impala_udf::StringVal::ptr, impala::DecimalAvgState::sum, and impala_udf::DecimalVal::val16.

void impala::AggregateFunctions::DecimalAvgRemove	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		StringVal *	dst
	)

static

Definition at line 240 of file aggregate-functions.cc.

References DecimalAvgAddOrRemove().

void impala::AggregateFunctions::DecimalAvgUpdate	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		StringVal *	dst
	)

static

Definition at line 235 of file aggregate-functions.cc.

References DecimalAvgAddOrRemove().

BigIntVal impala::AggregateFunctions::DenseRankGetValue	(	FunctionContext *	ctx,
		StringVal &	src
	)

static

Returns the result for DENSE_RANK and prepares the state for the next Update(). TODO: Implement DENSE_RANK with a single BigIntVal. Requires src can be modified, AggFnEvaluator would need to handle copying the src AnyVal back into the src slot.

Definition at line 1207 of file aggregate-functions.cc.

References impala::RankState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::RankState::rank.

void impala::AggregateFunctions::DenseRankUpdate	(	FunctionContext *	ctx,
		StringVal *	dst
	)

static

Update state for DENSE_RANK.

Definition at line 1190 of file aggregate-functions.cc.

template<typename T >

template void impala::AggregateFunctions::FirstValRewriteUpdate< DecimalVal >	(	FunctionContext *	,
		const T &	src,
		const BigIntVal &	,
		T *	dst
	)

static

Implements FIRST_VALUE for some windows that require rewrites during planning. The BigIntVal is unused by FirstValRewriteUpdate() (it is used by the AnalyticEvalNode).

Definition at line 1295 of file aggregate-functions.cc.

template<typename T >

void impala::AggregateFunctions::FirstValUpdate	(	FunctionContext *	ctx,
		const T &	src,
		T *	dst
	)

static

Implements FIRST_VALUE.

Definition at line 1270 of file aggregate-functions.cc.

References impala_udf::FunctionContext::impl(), impala::FunctionContextImpl::num_removes(), and impala::FunctionContextImpl::num_updates().

template<>

void impala::AggregateFunctions::FirstValUpdate	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

Definition at line 1281 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::FunctionContext::impl(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::FunctionContextImpl::num_removes(), impala::FunctionContextImpl::num_updates(), and impala_udf::StringVal::ptr.

template<typename T >

template StringVal impala::AggregateFunctions::HistogramFinalize< DecimalVal >	(	FunctionContext *	,
		const StringVal &	src
	)

static

Returns an equi-depth histogram computed from a sample of data produced via reservoir sampling. The result is a comma-separated list of up to 100 histogram bucket endpoints where each bucket contains the same number of elements. For example, "10, 50, 60, 100" would mean 25% of values are less than 10, 25% are between 10 and 50, etc.

Definition at line 957 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::NUM_BUCKETS, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, and impala::ReservoirSampleState< T >::samples.

uint64_t impala::AggregateFunctions::HllFinalEstimate	(	const uint8_t *	buckets,
		int32_t	num_buckets
	)

static

Utility method to compute the final result of an HLL estimation from num_buckets estimates.

Definition at line 1033 of file aggregate-functions.cc.

References HLL_LEN.

Referenced by HllFinalize().

BigIntVal impala::AggregateFunctions::HllFinalize	(	FunctionContext *	ctx,
		const StringVal &	src
	)

static

Definition at line 1069 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), HllFinalEstimate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::HllInit	(	FunctionContext *	ctx,
		StringVal *	slot
	)

static

Definition at line 998 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), HLL_LEN, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::HllMerge	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 1022 of file aggregate-functions.cc.

References HLL_LEN, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<typename T >

template void impala::AggregateFunctions::HllUpdate	(	FunctionContext *	,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 1007 of file aggregate-functions.cc.

References impala::HashUtil::FNV64_SEED, impala_udf::FunctionContext::GetArgType(), impala::AnyValUtil::Hash64(), impala::hash_value(), HLL_LEN, HLL_PRECISION, gen_ir_descriptions::idx, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::InitNull	(	FunctionContext *	,
		AnyVal *	dst
	)

static

Initializes dst to NULL.

Definition at line 62 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null.

Referenced by impala::Expr::InitBuiltinsDummy().

void impala::AggregateFunctions::InitNullString	(	FunctionContext *	c,
		StringVal *	dst
	)

static

Initializes dst to NULL and sets dst->ptr to NULL.

Definition at line 398 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<typename T >

void impala::AggregateFunctions::InitZero	(	FunctionContext *	,
		T *	dst
	)

static

Initializes dst to 0.

Definition at line 67 of file aggregate-functions.cc.

template<>

void impala::AggregateFunctions::InitZero	(	FunctionContext *	,
		DecimalVal *	dst
	)

Definition at line 73 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::DecimalVal::val16.

DoubleVal impala::AggregateFunctions::KnuthStddevFinalize	(	FunctionContext *	context,
		const StringVal &	val
	)

static

Calculates STDDEV, uses KnuthVar Init-Update-Merge functions.

Definition at line 1151 of file aggregate-functions.cc.

References impala::ComputeKnuthVariance(), impala::KnuthVarianceState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

DoubleVal impala::AggregateFunctions::KnuthStddevPopFinalize	(	FunctionContext *	context,
		const StringVal &	val
	)

static

Calculates the biased STDDEV, uses KnuthVar Init-Update-Merge functions.

Definition at line 1160 of file aggregate-functions.cc.

References impala::ComputeKnuthVariance(), impala::KnuthVarianceState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

DoubleVal impala::AggregateFunctions::KnuthVarFinalize	(	FunctionContext *	context,
		const StringVal &	val
	)

static

Definition at line 1134 of file aggregate-functions.cc.

References impala::ComputeKnuthVariance(), impala::KnuthVarianceState::count, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::KnuthVarInit	(	FunctionContext *	context,
		StringVal *	val
	)

static

Knuth's variance algorithm, more numerically stable than canonical stddev algorithms; reference implementation: http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Online_algorithm

Definition at line 1094 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::KnuthVarMerge	(	FunctionContext *	context,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 1115 of file aggregate-functions.cc.

References impala::KnuthVarianceState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::KnuthVarianceState::m2, impala::KnuthVarianceState::mean, and impala_udf::StringVal::ptr.

DoubleVal impala::AggregateFunctions::KnuthVarPopFinalize	(	FunctionContext *	context,
		const StringVal &	val
	)

static

Calculates the biased variance, uses KnuthVar Init-Update-Merge functions.

Definition at line 1142 of file aggregate-functions.cc.

References impala::ComputeKnuthVariance(), impala::KnuthVarianceState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<typename T >

template void impala::AggregateFunctions::KnuthVarUpdate	(	FunctionContext *	context,
		const T &	input,
		StringVal *	val
	)

static

Definition at line 1101 of file aggregate-functions.cc.

References impala::KnuthVarianceState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::KnuthVarianceState::m2, impala::KnuthVarianceState::mean, and impala_udf::StringVal::ptr.

template<typename T >

void impala::AggregateFunctions::LastValRemove	(	FunctionContext *	ctx,
		const T &	src,
		T *	dst
	)

static

Definition at line 1256 of file aggregate-functions.cc.

References impala_udf::FunctionContext::impl(), impala::FunctionContextImpl::num_removes(), and impala::FunctionContextImpl::num_updates().

template<>

void impala::AggregateFunctions::LastValRemove	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

Definition at line 1261 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::FunctionContext::impl(), impala_udf::AnyVal::is_null, impala::FunctionContextImpl::num_removes(), impala::FunctionContextImpl::num_updates(), and impala_udf::StringVal::ptr.

template<typename T >

void impala::AggregateFunctions::LastValUpdate	(	FunctionContext *	ctx,
		const T &	src,
		T *	dst
	)

static

Implements LAST_VALUE.

Definition at line 1232 of file aggregate-functions.cc.

template<>

void impala::AggregateFunctions::LastValUpdate	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

Definition at line 1237 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala_udf::FunctionContext::Reallocate().

template<typename T >

void impala::AggregateFunctions::Max	(	FunctionContext *	,
		const T &	src,
		T *	dst
	)

static

MaxUpdate/MaxMerge.

Definition at line 393 of file aggregate-functions.cc.

template<>

void impala::AggregateFunctions::Max	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

Definition at line 417 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::FunctionContext::Free(), impala::StringValue::FromStringVal(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<>

void impala::AggregateFunctions::Max	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst
	)

Definition at line 444 of file aggregate-functions.cc.

References impala_udf::FunctionContext::GetArgType(), impala_udf::AnyVal::is_null, impala_udf::FunctionContext::TypeDesc::precision, impala_udf::DecimalVal::val16, impala_udf::DecimalVal::val4, and impala_udf::DecimalVal::val8.

template<>

void impala::AggregateFunctions::Max	(	FunctionContext *	,
		const TimestampVal &	src,
		TimestampVal *	dst
	)

Definition at line 472 of file aggregate-functions.cc.

References impala::TimestampValue::FromTimestampVal(), and impala_udf::AnyVal::is_null.

template<typename T >

void impala::AggregateFunctions::Min	(	FunctionContext *	,
		const T &	src,
		T *	dst
	)

static

MinUpdate/MinMerge.

Definition at line 387 of file aggregate-functions.cc.

template<>

void impala::AggregateFunctions::Min	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	dst
	)

Definition at line 405 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::FunctionContext::Free(), impala::StringValue::FromStringVal(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<>

void impala::AggregateFunctions::Min	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst
	)

Definition at line 429 of file aggregate-functions.cc.

References impala_udf::FunctionContext::GetArgType(), impala_udf::AnyVal::is_null, impala_udf::FunctionContext::TypeDesc::precision, impala_udf::DecimalVal::val16, impala_udf::DecimalVal::val4, and impala_udf::DecimalVal::val8.

template<>

void impala::AggregateFunctions::Min	(	FunctionContext *	,
		const TimestampVal &	src,
		TimestampVal *	dst
	)

Definition at line 459 of file aggregate-functions.cc.

References impala::TimestampValue::FromTimestampVal(), and impala_udf::AnyVal::is_null.

template<typename T >

template void impala::AggregateFunctions::OffsetFnInit< DecimalVal >	(	FunctionContext *	,
		T *	dst
	)

static

OffsetFn*() implement LAG and LEAD. Init() sets the default value (the last constant parameter) as dst.

Definition at line 1301 of file aggregate-functions.cc.

References impala_udf::FunctionContext::GetArgType(), impala_udf::FunctionContext::GetConstantArg(), impala_udf::FunctionContext::GetNumArgs(), impala_udf::FunctionContext::IsArgConstant(), and impala_udf::FunctionContext::TypeDesc::type.

template<typename T >

template void impala::AggregateFunctions::OffsetFnUpdate< DecimalVal >	(	FunctionContext *	,
		const T &	src,
		const BigIntVal &	,
		const T &	,
		T *	dst
	)

static

Update() takes all the parameters to LEAD/LAG, including the integer offset and the default value, neither which are needed by Update(). (The offset is already used in the window for the analytic fn evaluation and the default value is set in Init().

Definition at line 1310 of file aggregate-functions.cc.

BigIntVal impala::AggregateFunctions::PcFinalize	(	FunctionContext *	c,
		const StringVal &	src
	)

static

Definition at line 712 of file aggregate-functions.cc.

References impala::DistinceEstimateFinalize(), impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::PcInit	(	FunctionContext *	c,
		StringVal *	slot
	)

static

Probabilistic Counting (PC), a distinct estimate algorithms. Probabilistic Counting with Stochastic Averaging (PCSA) is a variant of PC that runs faster and usually gets equally accurate results.

Definition at line 562 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::NUM_PC_BITMAPS, impala::PC_BITMAP_LENGTH, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::PcMerge	(	FunctionContext *	c,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 652 of file aggregate-functions.cc.

References impala::DistinctEstimateBitMapToString(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::NUM_PC_BITMAPS, impala::PC_BITMAP_LENGTH, impala_udf::StringVal::ptr, and VLOG_ROW.

BigIntVal impala::AggregateFunctions::PcsaFinalize	(	FunctionContext *	c,
		const StringVal &	src
	)

static

Definition at line 719 of file aggregate-functions.cc.

References impala::DistinceEstimateFinalize(), impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala::NUM_PC_BITMAPS, and impala_udf::StringVal::ptr.

template<typename T >

template void impala::AggregateFunctions::PcsaUpdate	(	FunctionContext *	,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 620 of file aggregate-functions.cc.

References impala_udf::FunctionContext::GetArgType(), impala::AnyValUtil::Hash(), impala::hash_value(), impala::NUM_PC_BITMAPS, impala::PC_BITMAP_LENGTH, impala_udf::StringVal::ptr, impala::SetDistinctEstimateBit(), and UNLIKELY.

template<typename T >

template void impala::AggregateFunctions::PcUpdate	(	FunctionContext *	,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 604 of file aggregate-functions.cc.

References impala_udf::FunctionContext::GetArgType(), impala::AnyValUtil::Hash(), impala::hash_value(), impala::NUM_PC_BITMAPS, impala::PC_BITMAP_LENGTH, impala_udf::StringVal::ptr, impala::SetDistinctEstimateBit(), and UNLIKELY.

BigIntVal impala::AggregateFunctions::RankFinalize	(	FunctionContext *	ctx,
		StringVal &	src
	)

static

Returns the result for RANK and DENSE_RANK and cleans up intermediate state in src.

Definition at line 1221 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::RankState::rank.

BigIntVal impala::AggregateFunctions::RankGetValue	(	FunctionContext *	ctx,
		StringVal &	src
	)

static

Returns the result for RANK and prepares the state for the next Update().

Definition at line 1192 of file aggregate-functions.cc.

References impala::RankState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala::RankState::rank.

void impala::AggregateFunctions::RankInit	(	FunctionContext *	ctx,
		StringVal *	slot
	)

static

Initializes the state for RANK and DENSE_RANK.

-----------------------—— Analytic Functions ---------------------------—— Analytic functions implement the UDA interface (except Merge(), Serialize()) and are used internally by the AnalyticEvalNode. Some analytic functions store intermediate state as a StringVal which is needed for multiple calls to Finalize(), so some fns also implement a (private) GetValue() method to just return the value. In that case, Finalize() is only called at the end to clean up.

Definition at line 1175 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::RankUpdate	(	FunctionContext *	ctx,
		StringVal *	dst
	)

static

Update state for RANK.

Definition at line 1183 of file aggregate-functions.cc.

References impala::RankState::count, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<typename T >

template StringVal impala::AggregateFunctions::ReservoirSampleFinalize< DecimalVal >	(	FunctionContext *	,
		const StringVal &	src
	)

static

Returns 20,000 unsorted samples as a list of comma-separated values.

Definition at line 938 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, and impala::ReservoirSampleState< T >::samples.

template<typename T >

template void impala::AggregateFunctions::ReservoirSampleInit< DecimalVal >	(	FunctionContext *	,
		StringVal *	slot
	)

static

Reservoir sampling produces a uniform random sample without knowing the total number of items. ReservoirSample{Init, Update, Merge, Serialize} implement distributed reservoir sampling. Samples are first collected locally via reservoir sampling in Update(), and then all local samples are merged together in Merge() using weights proportional to the size of the local input, using 'weighted' reservoir sampling. See the following references for more details: http://dl.acm.org/citation.cfm?id=1138834 http://gregable.com/2007/10/reservoir-sampling.html

Definition at line 792 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

template<typename T >

template void impala::AggregateFunctions::ReservoirSampleMerge< DecimalVal >	(	FunctionContext *	,
		const StringVal &	src,
		StringVal *	dst
	)

static

Definition at line 880 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::NUM_SAMPLES, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, impala::ReservoirSampleState< T >::samples, and impala::ReservoirSampleState< T >::source_size.

template<typename T >

template const StringVal impala::AggregateFunctions::ReservoirSampleSerialize< DecimalVal >	(	FunctionContext *	,
		const StringVal &	src
	)

static

Definition at line 819 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, impala::ReservoirSampleState< T >::samples, and impala::ReservoirSampleState< T >::source_size.

template<typename T >

template void impala::AggregateFunctions::ReservoirSampleUpdate	(	FunctionContext *	,
		const T &	src,
		StringVal *	dst
	)

static

Definition at line 802 of file aggregate-functions.cc.

References impala::ReservoirSampleState< T >::GetNext64(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala::NUM_SAMPLES, impala::ReservoirSampleState< T >::num_samples, impala_udf::StringVal::ptr, impala::ReservoirSampleState< T >::samples, and impala::ReservoirSampleState< T >::source_size.

StringVal impala::AggregateFunctions::StringConcatFinalize	(	FunctionContext *	ctx,
		const StringVal &	src
	)

static

Definition at line 533 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

void impala::AggregateFunctions::StringConcatMerge	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	result
	)

static

Definition at line 515 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala_udf::FunctionContext::Reallocate().

void impala::AggregateFunctions::StringConcatUpdate	(	FunctionContext *	ctx,
		const StringVal &	src,
		StringVal *	result
	)

static

String concat.

Definition at line 490 of file aggregate-functions.cc.

References impala::DEFAULT_STRING_CONCAT_DELIM.

void impala::AggregateFunctions::StringConcatUpdate	(	FunctionContext *	ctx,
		const StringVal &	src,
		const StringVal &	separator,
		StringVal *	result
	)

static

Definition at line 495 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Allocate(), impala::DEFAULT_STRING_CONCAT_DELIM, impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, and impala_udf::FunctionContext::Reallocate().

StringVal impala::AggregateFunctions::StringValGetValue	(	FunctionContext *	ctx,
		const StringVal &	src
	)

static

StringVal GetValue() function that returns a copy of src.

Definition at line 78 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, impala_udf::StringVal::len, and impala_udf::StringVal::ptr.

Referenced by StringValSerializeOrFinalize().

StringVal impala::AggregateFunctions::StringValSerializeOrFinalize	(	FunctionContext *	ctx,
		const StringVal &	src
	)

static

StringVal Serialize/Finalize function that copies and frees src.

Definition at line 86 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::AnyVal::is_null, impala_udf::StringVal::ptr, and StringValGetValue().

void impala::AggregateFunctions::SumDecimalAddOrSubtract	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst,
		bool	subtract = `false`
	)

static

Adds or or subtracts src from dst. Implements Update() and Remove().

Definition at line 362 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, impala_udf::DecimalVal::val16, impala_udf::DecimalVal::val4, and impala_udf::DecimalVal::val8.

Referenced by SumDecimalRemove(), and SumDecimalUpdate().

void impala::AggregateFunctions::SumDecimalMerge	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst
	)

static

Definition at line 379 of file aggregate-functions.cc.

References impala_udf::AnyVal::is_null, and impala_udf::DecimalVal::val16.

void impala::AggregateFunctions::SumDecimalRemove	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst
	)

static

Definition at line 353 of file aggregate-functions.cc.

References impala_udf::FunctionContext::impl(), impala::FunctionContextImpl::num_removes(), impala::FunctionContextImpl::num_updates(), and SumDecimalAddOrSubtract().

void impala::AggregateFunctions::SumDecimalUpdate	(	FunctionContext *	ctx,
		const DecimalVal &	src,
		DecimalVal *	dst
	)

static

Sum for decimals.

Definition at line 348 of file aggregate-functions.cc.

References SumDecimalAddOrSubtract().

template<typename SRC_VAL , typename DST_VAL >

template void impala::AggregateFunctions::SumRemove< DoubleVal, DoubleVal >	(	FunctionContext *	,
		const SRC_VAL &	src,
		DST_VAL *	dst
	)

static

Definition at line 335 of file aggregate-functions.cc.

References impala_udf::FunctionContext::impl(), impala::FunctionContextImpl::IncrementNumRemoves(), impala::FunctionContextImpl::num_removes(), and impala::FunctionContextImpl::num_updates().

template<typename SRC_VAL , typename DST_VAL >

template void impala::AggregateFunctions::SumUpdate< DoubleVal, DoubleVal >	(	FunctionContext *	,
		const SRC_VAL &	src,
		DST_VAL *	dst
	)

static

SumUpdate, SumMerge.

Definition at line 323 of file aggregate-functions.cc.

References impala_udf::FunctionContext::impl(), and impala::FunctionContextImpl::IncrementNumUpdates().

TimestampVal impala::AggregateFunctions::TimestampAvgFinalize	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 216 of file aggregate-functions.cc.

References impala_udf::FunctionContext::Free(), impala_udf::StringVal::ptr, and TimestampAvgGetValue().

TimestampVal impala::AggregateFunctions::TimestampAvgGetValue	(	FunctionContext *	ctx,
		const StringVal &	val
	)

static

Definition at line 206 of file aggregate-functions.cc.

References impala::AvgState::count, impala_udf::StringVal::ptr, impala::AvgState::sum, and impala::TimestampValue::ToTimestampVal().

Referenced by TimestampAvgFinalize().

void impala::AggregateFunctions::TimestampAvgRemove	(	FunctionContext *	ctx,
		const TimestampVal &	src,
		StringVal *	dst
	)

static

Definition at line 194 of file aggregate-functions.cc.

References impala::AvgState::count, impala::TimestampValue::FromTimestampVal(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, impala::AvgState::sum, and impala::TimestampValue::ToSubsecondUnixTime().

void impala::AggregateFunctions::TimestampAvgUpdate	(	FunctionContext *	ctx,
		const TimestampVal &	src,
		StringVal *	dst
	)

static

Avg for timestamp. Uses AvgInit() and AvgMerge().

Definition at line 183 of file aggregate-functions.cc.

References impala::AvgState::count, impala::TimestampValue::FromTimestampVal(), impala_udf::AnyVal::is_null, impala_udf::StringVal::len, impala_udf::StringVal::ptr, impala::AvgState::sum, and impala::TimestampValue::ToSubsecondUnixTime().

Member Data Documentation

const int impala::AggregateFunctions::HLL_LEN = 1024

static

Definition at line 175 of file aggregate-functions.h.

Referenced by HllFinalEstimate(), HllInit(), HllMerge(), and HllUpdate().

const int impala::AggregateFunctions::HLL_PRECISION = 10

static

Hyperloglog distinct estimate algorithm. See these papers for more details. 1) Hyperloglog: The analysis of a near-optimal cardinality estimation algorithm (2007) 2) HyperLogLog in Practice (paper from google with some improvements)

Definition at line 174 of file aggregate-functions.h.

Referenced by HllUpdate().

The documentation for this class was generated from the following files:

be/src/exprs/aggregate-functions.h
be/src/exprs/aggregate-functions.cc

Public Member Functions

Static Public Member Functions

Static Public Attributes

Detailed Description

Member Function Documentation

Member Data Documentation