A version of precise index that directly uses the integer values. More...

#include <idirekte.h>

Inheritance diagram for ibis::direkte:

Public Member Functions
virtual long	append (const char dt, const char df, uint32_t nnew)
	Append the index in df to the one in dt. More...

long	append (const ibis::direkte &tail)
	Append tail to this index. More...

long	append (const array_t< uint32_t > &ind)
	Append a list of integers. More...

virtual void	binBoundaries (std::vector< double > &) const
	The function binBoundaries and binWeights return bin boundaries and counts of each bin respectively. More...

virtual void	binWeights (std::vector< uint32_t > &) const

	direkte (const direkte &rhs)

	direkte (const ibis::column c, const char f=0)
	Constructing a new ibis::direkte object from base data in a file. More...

	direkte (const ibis::column c, ibis::fileManager::storage st)

	direkte (const ibis::column *c, uint32_t popu, uint32_t ntpl=0)
	Construct a dummy index. More...

	direkte (const ibis::column *c, uint32_t card, array_t< uint32_t > &ints)
	Construct an index from an integer array. More...

virtual index *	dup () const
	Duplicate the content of an index object.

virtual void	estimate (const ibis::qContinuousRange &expr, ibis::bitvector &lower, ibis::bitvector &upper) const
	Computes an approximation of hits as a pair of lower and upper bounds. More...

virtual uint32_t	estimate (const ibis::qContinuousRange &expr) const
	Returns an upper bound on the number of hits.

virtual void	estimate (const ibis::qDiscreteRange &expr, ibis::bitvector &lower, ibis::bitvector &upper) const
	Estimate the hits for discrete ranges, i.e., those translated from 'a IN (x, y, ..)'. More...

virtual uint32_t	estimate (const ibis::qDiscreteRange &expr) const

virtual double	estimateCost (const ibis::qContinuousRange &expr) const
	Estimate the cost of evaluating a range condition.

virtual double	estimateCost (const ibis::qDiscreteRange &expr) const
	Estimate the cost of evaluating a range condition.

virtual long	evaluate (const ibis::qContinuousRange &expr, ibis::bitvector &hits) const
	To evaluate the exact hits. More...

virtual long	evaluate (const ibis::qDiscreteRange &expr, ibis::bitvector &hits) const
	To evaluate the exact hits. More...

virtual long	getCumulativeDistribution (std::vector< double > &bds, std::vector< uint32_t > &cts) const
	Cumulative distribution of the data. More...

virtual long	getDistribution (std::vector< double > &bbs, std::vector< uint32_t > &cts) const
	Binned distribution of the data. More...

virtual double	getMax () const
	The maximum value recorded in the index.

virtual double	getMin () const
	The minimum value recorded in the index.

virtual double	getSum () const
	Compute the approximate sum of all the values indexed. More...

void	ints (array_t< uint32_t > &) const
	Convert the bitvector into integer values.

array_t< uint32_t > *	keys (const ibis::bitvector &mask) const
	Convert the bitvector mask into key values.

virtual const char *	name () const
	Returns the name of the index, similar to the function `type`, but returns a string instead. More...

virtual void	print (std::ostream &out) const
	The printing function.

virtual int	read (const char *name)
	Read index from the specified location.

virtual int	read (ibis::fileManager::storage *st)
	Reconstruct an index from a piece of consecutive memory.

int	remapKeys (const ibis::array_t< uint32_t > &)
	Change the key values to a new set of numbers. More...

virtual long	select (const ibis::qContinuousRange &, void *) const
	Evaluate the range condition and select values.

virtual long	select (const ibis::qContinuousRange &, void *, ibis::bitvector &) const
	Evaluate the range condition, select values, and record the positions.

virtual void	serialSizes (uint64_t &, uint64_t &, uint64_t &) const
	Compute the size of arrays that would be generated by the serializatioin function (write). More...

virtual void	speedTest (std::ostream &out) const
	Time some logical operations and print out their speed. More...

virtual INDEX_TYPE	type () const
	Returns an index type identifier.

virtual float	undecidable (const ibis::qContinuousRange &expr, ibis::bitvector &iffy) const
	Mark the position of the rows that can not be decided with this index. More...

virtual float	undecidable (const ibis::qDiscreteRange &expr, ibis::bitvector &iffy) const

virtual int	write (ibis::array_t< double > &, ibis::array_t< int64_t > &, ibis::array_t< uint32_t > &) const
	Save index to three arrays. Serialize the index in memory.

virtual int	write (const char *name) const
	Write the direct bitmap index to a file.

Public Member Functions inherited from ibis::index
void	addBins (uint32_t ib, uint32_t ie, ibis::bitvector &res) const
	Add the sum of `bits`[ib] through `bits`[ie-1] to `res`. More...

void	addBins (uint32_t ib, uint32_t ie, ibis::bitvector &res, const ibis::bitvector &tot) const
	Compute the sum of bit vectors [`ib`, `ie`). More...

virtual int	contractRange (ibis::qContinuousRange &) const

bool	empty () const
	The index object is considered empty if there is no bitmap or getNRows returns 0. More...

virtual void	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr, ibis::bitvector64 &lower, ibis::bitvector64 &upper) const
	Estimate the pairs for the range join operator.

virtual void	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr, const ibis::bitvector &mask, ibis::bitvector64 &lower, ibis::bitvector64 &upper) const
	Estimate the pairs for the range join operator. More...

virtual void	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr, const ibis::bitvector &mask, const ibis::qRange const range1, const ibis::qRange const range2, ibis::bitvector64 &lower, ibis::bitvector64 &upper) const

virtual int64_t	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr) const
	Estimate an upper bound for the number of pairs.

virtual int64_t	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr, const ibis::bitvector &mask) const
	Estimate an upper bound for the number of pairs produced from marked records. More...

virtual int64_t	estimate (const ibis::index &idx2, const ibis::deprecatedJoin &expr, const ibis::bitvector &mask, const ibis::qRange const range1, const ibis::qRange const range2) const

virtual void	estimate (const ibis::deprecatedJoin &expr, const ibis::bitvector &mask, const ibis::qRange const range1, const ibis::qRange const range2, ibis::bitvector64 &lower, ibis::bitvector64 &upper) const
	Evaluating a join condition with one (likely composite) index.

virtual int64_t	estimate (const ibis::deprecatedJoin &expr, const ibis::bitvector &mask, const ibis::qRange const range1, const ibis::qRange const range2) const

virtual int	expandRange (ibis::qContinuousRange &) const
	The functions expandRange and contractRange expands or contracts the boundaries of a range condition so that the new range will have exact answers using the function estimate. More...

virtual const ibis::bitvector *	getBitvector (uint32_t i) const
	Return a pointer to the ith bitvector used in the index (may be 0).

uint32_t	getNRows () const
	Return the number of rows represented by this object.

virtual uint32_t	numBitvectors () const
	Returns the number of bit vectors used by the index.

float	sizeInBytes () const
	Estiamte the size of this index object measured in bytes. More...

void	sumBins (uint32_t ib, uint32_t ie, ibis::bitvector &res) const
	Sum up `bits[ib:ie-1]` and place the result in res. More...

void	sumBins (uint32_t ib, uint32_t ie, ibis::bitvector &res, uint32_t ib0, uint32_t ie0) const
	Compute a new sum for bit vectors [ib, ie) by taking advantage of the old sum for bitvectors [ib0, ie0). More...

void	sumBins (uint32_t ib, uint32_t ie, ibis::bitvector &res, uint32_t *buf) const
	Sum up `bits[ib:ie-1]` and place the result in res. More...

void	sumBins (const ibis::array_t< uint32_t > &, ibis::bitvector &) const
	Sum up the bits in in the specified bins.

virtual	~index ()
	The destructor.

Protected Member Functions
template<typename T >
int	construct (const char *f)

template<typename T >
int	construct0 (const char *f)

virtual size_t	getSerialSize () const throw ()
	Estimate the size of the index file. More...

void	locate (const ibis::qContinuousRange &expr, uint32_t &hit0, uint32_t &hit1) const

direkte &	operator= (const direkte &)

Protected Member Functions inherited from ibis::index
virtual void	activate () const
	Regenerate all bitvectors from the underlying storage. More...

virtual void	activate (uint32_t i) const
	Regenerate the ith bitvector from the underlying storage.

virtual void	activate (uint32_t i, uint32_t j) const
	Regenerate bitvectors i (inclusive) through j (exclusive) from the underlying storage. More...

virtual void	clear ()
	Clear the existing content. More...

void	computeMinMax (const char *f, double &min, double &max) const

void	dataFileName (std::string &name, const char *f=0) const
	Generate data file name from "f". More...

	index (const ibis::column *c=0)
	Default constructor. More...

	index (const ibis::column c, ibis::fileManager::storage s)
	Constructor with a storage object. More...

	index (const index &)
	Copy constructor.

void	indexFileName (std::string &name, const char *f=0) const
	Generates index file name from "f". More...

void	initBitmaps (int fdes)
	Prepare the bitmaps using the given file descriptor. More...

void	initBitmaps (ibis::fileManager::storage *st)
	Prepare bitmaps from the given storage object. More...

void	initBitmaps (uint32_t *st)
	Prepare bitmaps from the given raw pointer. More...

void	initBitmaps (void *ctx, FastBitReadBitmaps rd)
	Prepare bitmaps from the user provided function pointer and context. More...

int	initOffsets (int64_t *, size_t)
	Initialize the offsets from the given data array. More...

int	initOffsets (int fdes, const char offsize, size_t start, uint32_t nobs)
	Read in the offset array. More...

int	initOffsets (ibis::fileManager::storage *st, size_t start, uint32_t nobs)
	Regenerate the offsets array from the given storage object. More...

void	mapValues (const char *f, VMap &bmap) const
	Map the positions of each individual value. More...

void	mapValues (const char *f, histogram &hist, uint32_t count=0) const
	Generate a histogram. More...

index &	operator= (const index &)
	Assignment operator.

void	optionalUnpack (array_t< ibis::bitvector * > &bits, const char *opt)
	A function to decide whether to uncompress the bitvectors. More...

Additional Inherited Members
Public Types inherited from ibis::index
typedef std::map< double, uint32_t >	histogram

enum	INDEX_TYPE { BINNING =0, RANGE, MESA, AMBIT, PALE, PACK, ZONE, RELIC, ROSTER, SKIVE, FADE, SBIAD, SAPID, EGALE, MOINS, ENTRE, BAK, BAK2, KEYWORDS, MESH, BAND, DIREKTE, GENERIC, BYLT, FUZZ, ZONA, FUGE, SLICE, EXTERN }
	The integer values of this enum type are used in the index files to differentiate the indexes. More...

typedef std::map< double, ibis::bitvector * >	VMap

Static Public Member Functions inherited from ibis::index
static void	addBits (const array_t< bitvector * > &bits, uint32_t ib, uint32_t ie, ibis::bitvector &res)
	Add the `pile`[ib:ie-1] to `res`. More...

static index *	create (const column c, const char name=0, const char *spec=0, int inEntirety=0)
	Index factory. More...

static void	divideCounts (array_t< uint32_t > &bounds, const array_t< uint32_t > &cnt)
	Determine how to split the array `cnt`, so that each group has roughly the same total value. More...

static bool	isIndex (const char *f, INDEX_TYPE t)
	Is the named file an index file? Read the header of the named file to determine if it contains an index of the specified type. More...

template<typename E >
static void	mapValues (const array_t< E > &val, VMap &bmap)

template<typename E >
static void	mapValues (const array_t< E > &val, histogram &hist, uint32_t count=0)

template<typename E >
static void	mapValues (const array_t< E > &val, array_t< E > &bounds, std::vector< uint32_t > &cnts)

template<typename E1 , typename E2 >
static void	mapValues (const array_t< E1 > &val1, const array_t< E2 > &val2, array_t< E1 > &bnd1, array_t< E2 > &bnd2, std::vector< uint32_t > &cnts)
	Compute a two-dimensional histogram. More...

static void	printHeader (std::ostream &, const char *)

static void	setBases (array_t< uint32_t > &bases, uint32_t card, uint32_t nbase=2)
	Fill the array bases with the values that cover the range [0, card). More...

static void	sumBits (const array_t< bitvector * > &bits, uint32_t ib, uint32_t ie, ibis::bitvector &res)
	Sum up `pile`[ib:ie-1] and place the result in `res`. More...

static void	sumBits (const array_t< bitvector * > &bits, const ibis::bitvector &tot, uint32_t ib, uint32_t ie, ibis::bitvector &res)
	Sum up `pile`[ib:ie-1] and add the result to `res`. More...

Static Protected Member Functions inherited from ibis::index
static void	indexFileName (std::string &name, const ibis::column col1, const ibis::column col2, const char *f=0)
	Generate the index file name for the composite index fromed on two columns. More...

Protected Attributes inherited from ibis::index
array_t< ibis::bitvector * >	bits
	A list of bitvectors.

bitmapReader *	breader
	The functor to read serialized bitmaps from a more complex source.

const ibis::column *	col
	Pointer to the column this index is for.

const char *	fname
	The name of the file containing the index.

uint32_t	nrows
	The number of rows represented by the index. More...

array_t< int32_t >	offset32
	Starting positions of the bitvectors.

array_t< int64_t >	offset64
	Starting positions of the bitvectors. More...

ibis::fileManager::storage *	str
	The underlying storage. More...

Detailed Description

A version of precise index that directly uses the integer values.

It can avoid some intemdiate steps during index building and query answering. However, this class can only be used with integer column with nonnegative values. Ideally, the values should start with 0, and only use small positive integers.

Constructor & Destructor Documentation

ibis::direkte::direkte	(	const ibis::column *	c,
		const char *	f = `0`
	)

Constructing a new ibis::direkte object from base data in a file.

Both arguments are expected to be valid pointers.

References ibis::index::bits, ibis::CATEGORY, ibis::index::col, ibis::index::dataFileName(), ibis::DOUBLE, ibis::FLOAT, ibis::column::fullname(), ibis::INT, ibis::LONG, ibis::column::lowerBound(), ibis::column::name(), print(), read(), ibis::SHORT, ibis::array_t< T >::size(), ibis::TEXT, ibis::column::type(), ibis::TYPESTRING, ibis::UBYTE, ibis::UINT, ibis::ULONG, ibis::column::upperBound(), and ibis::USHORT.

ibis::direkte::direkte	(	const ibis::column *	c,
		uint32_t	popu,
		uint32_t	ntpl = `0`
	)

Construct a dummy index.

All rows are marked as having the same value with position popu. This creates an index with (popu+1) bit vectors, with the last one set to all 1s and the rest to be empty.

References ibis::index::bits, ibis::index::clear(), ibis::index::col, ibis::column::fullname(), ibis::part::nRows(), ibis::index::nrows, print(), and ibis::array_t< T >::resize().

ibis::direkte::direkte	(	const ibis::column *	c,
		uint32_t	card,
		array_t< uint32_t > &	ind
	)

Construct an index from an integer array.

The values in the array ind are assumed to be between 0 and card-1. All values outside of this range are ignored.

References ibis::index::bits, ibis::index::clear(), ibis::index::col, ibis::column::fullname(), ibis::index::nrows, print(), ibis::array_t< T >::resize(), and ibis::array_t< T >::size().

Member Function Documentation

long ibis::direkte::append	(	const char *	dt,
		const char *	df,
		uint32_t	nnew
	)

virtual

Append the index in df to the one in dt.

If the index in df exists, then it will be used, otherwise it simply creates a new index using the data in dt.

Reimplemented from ibis::index.

References ibis::fileManager::storage::begin(), ibis::index::bits, ibis::CATEGORY, ibis::util::clear(), ibis::index::DIREKTE, ibis::fileManager::getFile(), ibis::fileManager::instance(), ibis::INT, ibis::LONG, ibis::index::nrows, ibis::index::offset32, ibis::index::offset64, ibis::SHORT, ibis::array_t< T >::size(), ibis::index::str, ibis::fileManager::storage::swap(), ibis::TYPESTRING, ibis::UBYTE, ibis::UINT, ibis::ULONG, and ibis::USHORT.

Referenced by ibis::category::append(), and ibis::category::fillIndex().

long ibis::direkte::append ( const ibis::direkte & tail )

Append tail to this index.

The incoming index must be for the same column as this one.

References ibis::index::activate(), ibis::index::bits, ibis::index::col, ibis::index::nrows, ibis::bitvector::set(), and ibis::array_t< T >::size().

long ibis::direkte::append ( const array_t< uint32_t > & ind )

Append a list of integers.

The integers are treated as bin numbers. This function is primarily used by ibis::category::append.

References ibis::array_t< T >::push_back(), and ibis::array_t< T >::size().

void ibis::direkte::binBoundaries ( std::vector< double > & ) const

virtual

The function binBoundaries and binWeights return bin boundaries and counts of each bin respectively.

Reimplemented from ibis::index.

void ibis::direkte::estimate	(	const ibis::qContinuousRange &	,
		ibis::bitvector &	lower,
		ibis::bitvector &	upper
	)		const

virtual

Computes an approximation of hits as a pair of lower and upper bounds.

Parameters

expr	the query expression to be evaluated.
lower	a bitvector marking a subset of the hits. All rows marked with one (1) are definitely hits.
upper	a bitvector marking a superset of the hits. All hits are marked with one, but some of the rows marked one may not be hits. If the variable upper is empty, the variable lower is assumed to contain the exact answer.

Reimplemented from ibis::index.

References ibis::bitvector::clear().

void ibis::direkte::estimate	(	const ibis::qDiscreteRange &	expr,
		ibis::bitvector &	lower,
		ibis::bitvector &	upper
	)		const

virtual

Estimate the hits for discrete ranges, i.e., those translated from 'a IN (x, y, ..)'.

A trivial implementation to indicate the index can not determine any row.

Reimplemented from ibis::index.

References ibis::bitvector::clear(), ibis::qDiscreteRange::getValues(), ibis::bitvector::set(), and ibis::array_t< T >::size().

long ibis::direkte::evaluate	(	const ibis::qContinuousRange &	expr,
		ibis::bitvector &	hits
	)		const

virtual

To evaluate the exact hits.

On success, return the number of hits, otherwise a negative value is returned.

Implements ibis::index.

References ibis::bitvector::cnt().

long ibis::direkte::evaluate	(	const ibis::qDiscreteRange &	,
		ibis::bitvector &
	)		const

virtual

To evaluate the exact hits.

On success, return the number of hits, otherwise a negative value is returned.

Reimplemented from ibis::index.

References ibis::bitvector::cnt(), ibis::qDiscreteRange::getValues(), ibis::bitvector::set(), and ibis::array_t< T >::size().

long ibis::direkte::getCumulativeDistribution	(	std::vector< double > &	bds,
		std::vector< uint32_t > &	cts
	)		const

virtual

Cumulative distribution of the data.

A brute-force approach to get an accurate cumulative distribution.

Reimplemented from ibis::index.

long ibis::direkte::getDistribution	(	std::vector< double > &	bbs,
		std::vector< uint32_t > &	cts
	)		const

virtual

Binned distribution of the data.

A brute-force approach to get an accurate distribution.

Reimplemented from ibis::index.

size_t ibis::direkte::getSerialSize	(	)	const
throw	(
	)

protectedvirtual

Estimate the size of the index file.

The index file contains primarily the bitmaps.

Reimplemented from ibis::index.

double ibis::direkte::getSum ( ) const

virtual

Compute the approximate sum of all the values indexed.

If it decides that computing the sum directly from the vertical partition is more efficient, it will return NaN immediately.

Reimplemented from ibis::index.

virtual const char* ibis::direkte::name ( ) const

inlinevirtual

Returns the name of the index, similar to the function type, but returns a string instead.

Implements ibis::index.

int ibis::direkte::remapKeys ( const ibis::array_t< uint32_t > & o2n )

Change the key values to a new set of numbers.

This is used after a categorical value column changes it dictionary and we need to reshuffle the bitmaps but not the actual content in any bitmap. The incoming argument is expected to be an array of exactly the same number of elements as the number of bitmaps in this index.

Return the number of bit vectors after successfully remapped the keys. Otherwise return a negative number.

References ibis::array_t< T >::clear(), ibis::util::clear(), ibis::array_t< T >::copy(), ibis::fileManager::flushFile(), ibis::fileManager::instance(), ibis::array_t< T >::size(), ibis::array_t< T >::swap(), and ibis::util::write().

Referenced by ibis::category::setDictionary().

void ibis::direkte::serialSizes	(	uint64_t &	,
		uint64_t &	,
		uint64_t &
	)		const

virtual

Compute the size of arrays that would be generated by the serializatioin function (write).

Implements ibis::index.

virtual void ibis::direkte::speedTest ( std::ostream & out ) const

inlinevirtual

Time some logical operations and print out their speed.

This version does nothing.

Reimplemented from ibis::index.

virtual float ibis::direkte::undecidable	(	const ibis::qContinuousRange &	expr,
		ibis::bitvector &	iffy
	)		const

inlinevirtual

Mark the position of the rows that can not be decided with this index.

Parameters

expr	the range conditions to be evaluated.
iffy	the bitvector marking the positions of rows that can not be decided using the index. Return value is the expected fraction of undecided rows that might satisfy the range conditions.

Reimplemented from ibis::index.

References ibis::bitvector::clear().

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

idirekte.h
idirekte.cpp



	Contact us Disclaimers FastBit source code FastBit mailing list archive

Public Member Functions

Protected Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation