hll.h File Reference

HyperLogLog functions, based on code by Ivan Vitjuk https://github.com/ivitjuk/libhll under an ISC License. More...

#include "hashfunctions.h"

Include dependency graph for hll.h:

Go to the source code of this file.

Data Structures
struct	hll_estimate_s

Typedefs
typedef struct hll_s	hll_t
typedef struct hll_estimate_s	hll_estimate_t
typedef uint64_t(*	hll_hash_function_t) (const char *, size_t)

Functions
hll_t *	hll_create (size_t bucket_bits)
void	hll_reset (hll_t *hll)
void	hll_release (hll_t *hll)
void	hll_add (const hll_t *hll, const char *data, size_t data_len)
int	hll_merge (const hll_t *hll1, const hll_t *hll2)
int	hll_set_hash_function (hll_t *hll, hll_hash_function_t hash_function)
int	hll_get_estimate (const hll_t *hll, hll_estimate_t *estimate)

Detailed Description

HyperLogLog functions, based on code by Ivan Vitjuk https://github.com/ivitjuk/libhll under an ISC License.

Typedef Documentation

hll_estimate_t

typedef struct hll_estimate_s hll_estimate_t

Estimation result type

hll_hash_function_t

typedef uint64_t(* hll_hash_function_t) (const char *, size_t)

Hash function type

Even though hash funtion is expected to return a 64bit value, only 32 bits of entropy will be used.

Function Documentation

hll_create()

hll_t* hll_create

(

size_t

bucket_bits

)

Create HLL data structure

Parameters

bucket_bits

- Number of bits to use for the buckets. Actual number of buckets will be 2^bucket_bits. Must be 4 <= bucket_bits <= 16.

Returns

HLL data type or 0 on error. Error can be memory allocation failure or invalid bucket_bits value.

hll_reset()

void hll_reset

(

hll_t *

hll

)

Reset state of the estimator

Parameters

hll	- HLL data type

hll_release()

void hll_release

(

hll_t *

hll

)

Release HLL type previously allocated with hll_create().

Parameters

hll	- HLL data type

hll_add()

void hll_add	(	const hll_t *	hll,
		const char *	data,
		size_t	data_len
	)

Add a sample to the HLL estimator

Parameters

hll	- HLL data type
data	- Sample to be added to the estimator (underlying data type is not important)
data_len	- Lenght of the data sample in bytes

hll_merge()

int hll_merge	(	const hll_t *	hll1,
		const hll_t *	hll2
	)

Merge data from two HLLs

Data from hll2 will be merged into hll2

Parameters

hll1	- First HLL data type
hll2	- Second HLL data type

Returns

1 on success, 0 on failure. Fails when number of buckets are not compatible.

hll_get_estimate()

int hll_get_estimate	(	const hll_t *	hll,
		hll_estimate_t *	estimate
	)

Get the estimated cardinality based on the data added to the estimator

Parameters

hll	- HLL data type
estimate	- Result of the estimation

Returns

1 on success, 0 on failure. Fails only on NULL input parameters.