ddidderr/zstd-rs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[zdict] Make COVER the default algorithm

Browse Source
This commit is contained in:
Nick Terrell
2017-06-26 21:09:22 -07:00
parent ef269c1b68
commit 5b7fd7c422
8 changed files with 182 additions and 196 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/dictBuilder/zdict.h
+93 -86
View File
@@ -36,18 +36,20 @@ extern "C" {
#endif
/*! ZDICT_trainFromBuffer() :
Train a dictionary from an array of samples.
Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
The resulting dictionary will be saved into `dictBuffer`.
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
or an error code, which can be tested with ZDICT_isError().
Tips : In general, a reasonable dictionary has a size of ~ 100 KB.
It's obviously possible to target smaller or larger ones, just by specifying different `dictBufferCapacity`.
In general, it's recommended to provide a few thousands samples, but this can vary a lot.
It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
*/
/*! ZDICT_trainFromBuffer():
* Train a dictionary from an array of samples.
* Uses ZDICT_optimizeTrainFromBuffer_cover() single-threaded, with d=8 and steps=4.
* Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
* supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
* The resulting dictionary will be saved into `dictBuffer`.
* @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
* or an error code, which can be tested with ZDICT_isError().
* Note: ZDICT_trainFromBuffer() requires about 9 bytes of memory for each input byte.
* Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
* It's obviously possible to target smaller or larger ones, just by specifying different `dictBufferCapacity`.
* In general, it's recommended to provide a few thousands samples, but this can vary a lot.
* It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
*/
ZDICTLIB_API size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples);
@@ -69,94 +71,78 @@ ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode);
* ==================================================================================== */
typedef struct {
unsigned selectivityLevel; /* 0 means default; larger => select more => larger dictionary */
int compressionLevel; /* 0 means default; target a specific zstd compression level */
unsigned notificationLevel; /* Write to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
unsigned dictID; /* 0 means auto mode (32-bits random value); other : force dictID value */
unsigned reserved[2]; /* reserved space for future parameters */
} ZDICT_params_t;
/*! ZDICT_trainFromBuffer_advanced() :
Same as ZDICT_trainFromBuffer() with control over more parameters.
`parameters` is optional and can be provided with values set to 0 to mean "default".
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferSize`),
or an error code, which can be tested by ZDICT_isError().
note : ZDICT_trainFromBuffer_advanced() will send notifications into stderr if instructed to, using notificationLevel>0.
*/
ZDICTLIB_API size_t ZDICT_trainFromBuffer_advanced(void* dictBuffer, size_t dictBufferCapacity,
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
ZDICT_params_t parameters);
/*! COVER_params_t :
For all values 0 means default.
k and d are the only required parameters.
*/
/*! ZDICT_cover_params_t:
* For all values 0 means default.
* k and d are the only required parameters.
*/
typedef struct {
unsigned k; /* Segment size : constraint: 0 < k : Reasonable range [16, 2048+] */
unsigned d; /* dmer size : constraint: 0 < d <= k : Reasonable range [6, 16] */
unsigned steps; /* Number of steps : Only used for optimization : 0 means default (32) : Higher means more parameters checked */
unsigned nbThreads; /* Number of threads : constraint: 0 < nbThreads : 1 means single-threaded : Only used for optimization : Ignored if ZSTD_MULTITHREAD is not defined */
unsigned notificationLevel; /* Write to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */
unsigned dictID; /* 0 means auto mode (32-bits random value); other : force dictID value */
int compressionLevel; /* 0 means default; target a specific zstd compression level */
} COVER_params_t;
ZDICT_params_t zParams;
} ZDICT_cover_params_t;
/*! COVER_trainFromBuffer() :
Train a dictionary from an array of samples using the COVER algorithm.
Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
The resulting dictionary will be saved into `dictBuffer`.
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
or an error code, which can be tested with ZDICT_isError().
Note : COVER_trainFromBuffer() requires about 9 bytes of memory for each input byte.
Tips : In general, a reasonable dictionary has a size of ~ 100 KB.
It's obviously possible to target smaller or larger ones, just by specifying different `dictBufferCapacity`.
In general, it's recommended to provide a few thousands samples, but this can vary a lot.
It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
*/
ZDICTLIB_API size_t COVER_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
COVER_params_t parameters);
/*! ZDICT_trainFromBuffer_cover():
* Train a dictionary from an array of samples using the COVER algorithm.
* Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
* supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
* The resulting dictionary will be saved into `dictBuffer`.
* @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
* or an error code, which can be tested with ZDICT_isError().
* Note: ZDICT_trainFromBuffer_cover() requires about 9 bytes of memory for each input byte.
* Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
* It's obviously possible to target smaller or larger ones, just by specifying different `dictBufferCapacity`.
* In general, it's recommended to provide a few thousands samples, but this can vary a lot.
* It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
*/
ZDICTLIB_API size_t ZDICT_trainFromBuffer_cover(
void *dictBuffer, size_t dictBufferCapacity, const void *samplesBuffer,
const size_t *samplesSizes, unsigned nbSamples,
ZDICT_cover_params_t parameters);
/*! COVER_optimizeTrainFromBuffer() :
The same requirements as above hold for all the parameters except `parameters`.
This function tries many parameter combinations and picks the best parameters.
`*parameters` is filled with the best parameters found, and the dictionary
constructed with those parameters is stored in `dictBuffer`.
/*! ZDICT_optimizeTrainFromBuffer_cover():
* The same requirements as above hold for all the parameters except `parameters`.
* This function tries many parameter combinations and picks the best parameters.
* `*parameters` is filled with the best parameters found, and the dictionary
* constructed with those parameters is stored in `dictBuffer`.
*
* All of the parameters d, k, steps are optional.
* If d is non-zero then we don't check multiple values of d, otherwise we check d = {6, 8, 10, 12, 14, 16}.
* if steps is zero it defaults to its default value.
* If k is non-zero then we don't check multiple values of k, otherwise we check steps values in [16, 2048].
*
* @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
* or an error code, which can be tested with ZDICT_isError().
* On success `*parameters` contains the parameters selected.
* Note: ZDICT_optimizeTrainFromBuffer_cover() requires about 8 bytes of memory for each input byte and additionally another 5 bytes of memory for each byte of memory for each thread.
*/
ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_cover(
void *dictBuffer, size_t dictBufferCapacity, const void *samplesBuffer,
const size_t *samplesSizes, unsigned nbSamples,
ZDICT_cover_params_t *parameters);
All of the parameters d, k, steps are optional.
If d is non-zero then we don't check multiple values of d, otherwise we check d = {6, 8, 10, 12, 14, 16}.
if steps is zero it defaults to its default value.
If k is non-zero then we don't check multiple values of k, otherwise we check steps values in [16, 2048].
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
or an error code, which can be tested with ZDICT_isError().
On success `*parameters` contains the parameters selected.
Note : COVER_optimizeTrainFromBuffer() requires about 8 bytes of memory for each input byte and additionally another 5 bytes of memory for each byte of memory for each thread.
*/
ZDICTLIB_API size_t COVER_optimizeTrainFromBuffer(void* dictBuffer, size_t dictBufferCapacity,
const void* samplesBuffer, const size_t *samplesSizes, unsigned nbSamples,
COVER_params_t *parameters);
/*! ZDICT_finalizeDictionary() :
Given a custom content as a basis for dictionary, and a set of samples,
finalize dictionary by adding headers and statistics.
Samples must be stored concatenated in a flat buffer `samplesBuffer`,
supplied with an array of sizes `samplesSizes`, providing the size of each sample in order.
dictContentSize must be >= ZDICT_CONTENTSIZE_MIN bytes.
maxDictSize must be >= dictContentSize, and must be >= ZDICT_DICTSIZE_MIN bytes.
@return : size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`),
or an error code, which can be tested by ZDICT_isError().
note : ZDICT_finalizeDictionary() will push notifications into stderr if instructed to, using notificationLevel>0.
note 2 : dictBuffer and dictContent can overlap
*/
/*! ZDICT_finalizeDictionary():
* Given a custom content as a basis for dictionary, and a set of samples,
* finalize dictionary by adding headers and statistics.
*
* Samples must be stored concatenated in a flat buffer `samplesBuffer`,
* supplied with an array of sizes `samplesSizes`, providing the size of each sample in order.
*
* dictContentSize must be >= ZDICT_CONTENTSIZE_MIN bytes.
* maxDictSize must be >= dictContentSize, and must be >= ZDICT_DICTSIZE_MIN bytes.
*
* @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`),
* or an error code, which can be tested by ZDICT_isError().
* Note: ZDICT_finalizeDictionary() will push notifications into stderr if instructed to, using notificationLevel>0.
* Note 2: dictBuffer and dictContent can overlap
*/
#define ZDICT_CONTENTSIZE_MIN 128
#define ZDICT_DICTSIZE_MIN 256
ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBufferCapacity,
@@ -164,7 +150,28 @@ ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBuffer
const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples,
ZDICT_params_t parameters);
typedef struct {
unsigned selectivityLevel; /* 0 means default; larger => select more => larger dictionary */
ZDICT_params_t zParams;
} ZDICT_legacy_params_t;
/*! ZDICT_trainFromBuffer_legacy():
* Train a dictionary from an array of samples.
* Samples must be stored concatenated in a single flat buffer `samplesBuffer`,
* supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order.
* The resulting dictionary will be saved into `dictBuffer`.
* `parameters` is optional and can be provided with values set to 0 to mean "default".
* @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`)
* or an error code, which can be tested with ZDICT_isError().
* Tips: In general, a reasonable dictionary has a size of ~ 100 KB.
* It's obviously possible to target smaller or larger ones, just by specifying different `dictBufferCapacity`.
* In general, it's recommended to provide a few thousands samples, but this can vary a lot.
* It's recommended that total size of all samples be about ~x100 times the target size of dictionary.
* Note: ZDICT_trainFromBuffer_legacy() will send notifications into stderr if instructed to, using notificationLevel>0.
*/
ZDICTLIB_API size_t ZDICT_trainFromBuffer_legacy(
void *dictBuffer, size_t dictBufferCapacity, const void *samplesBuffer,
const size_t *samplesSizes, unsigned nbSamples, ZDICT_legacy_params_t parameters);
/* Deprecation warnings */
/* It is generally possible to disable deprecation warnings from compiler,