1/* 2 * Copyright (c) 2016-present, Yann Collet, Facebook, Inc. 3 * All rights reserved. 4 * 5 * This source code is licensed under both the BSD-style license (found in the 6 * LICENSE file in the root directory of this source tree) and the GPLv2 (found 7 * in the COPYING file in the root directory of this source tree). 8 * You may select, at your option, one of the above-listed licenses. 9 */ 10 11#ifndef DICTBUILDER_H_001 12#define DICTBUILDER_H_001 13 14#if defined (__cplusplus) 15extern "C" { 16#endif 17 18 19/*====== Dependencies ======*/ 20#include <stddef.h> /* size_t */ 21 22 23/* ===== ZDICTLIB_API : control library symbols visibility ===== */ 24#ifndef ZDICTLIB_VISIBILITY 25# if defined(__GNUC__) && (__GNUC__ >= 4) 26# define ZDICTLIB_VISIBILITY __attribute__ ((visibility ("default"))) 27# else 28# define ZDICTLIB_VISIBILITY 29# endif 30#endif 31#if defined(ZSTD_DLL_EXPORT) && (ZSTD_DLL_EXPORT==1) 32# define ZDICTLIB_API __declspec(dllexport) ZDICTLIB_VISIBILITY 33#elif defined(ZSTD_DLL_IMPORT) && (ZSTD_DLL_IMPORT==1) 34# define ZDICTLIB_API __declspec(dllimport) ZDICTLIB_VISIBILITY /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/ 35#else 36# define ZDICTLIB_API ZDICTLIB_VISIBILITY 37#endif 38 39 40/*! ZDICT_trainFromBuffer(): 41 * Train a dictionary from an array of samples. 42 * Redirect towards ZDICT_optimizeTrainFromBuffer_cover() single-threaded, with d=8 and steps=4. 43 * Samples must be stored concatenated in a single flat buffer `samplesBuffer`, 44 * supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order. 45 * The resulting dictionary will be saved into `dictBuffer`. 46 * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`) 47 * or an error code, which can be tested with ZDICT_isError(). 48 * Note: ZDICT_trainFromBuffer() requires about 9 bytes of memory for each input byte. 49 * Tips: In general, a reasonable dictionary has a size of ~ 100 KB. 50 * It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`. 51 * In general, it's recommended to provide a few thousands samples, though this can vary a lot. 52 * It's recommended that total size of all samples be about ~x100 times the target size of dictionary. 53 */ 54ZDICTLIB_API size_t ZDICT_trainFromBuffer(void* dictBuffer, size_t dictBufferCapacity, 55 const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples); 56 57 58/*====== Helper functions ======*/ 59ZDICTLIB_API unsigned ZDICT_getDictID(const void* dictBuffer, size_t dictSize); /**< extracts dictID; @return zero if error (not a valid dictionary) */ 60ZDICTLIB_API unsigned ZDICT_isError(size_t errorCode); 61ZDICTLIB_API const char* ZDICT_getErrorName(size_t errorCode); 62 63 64 65#ifdef ZDICT_STATIC_LINKING_ONLY 66 67/* ==================================================================================== 68 * The definitions in this section are considered experimental. 69 * They should never be used with a dynamic library, as they may change in the future. 70 * They are provided for advanced usages. 71 * Use them only in association with static linking. 72 * ==================================================================================== */ 73 74typedef struct { 75 int compressionLevel; /* optimize for a specific zstd compression level; 0 means default */ 76 unsigned notificationLevel; /* Write log to stderr; 0 = none (default); 1 = errors; 2 = progression; 3 = details; 4 = debug; */ 77 unsigned dictID; /* force dictID value; 0 means auto mode (32-bits random value) */ 78} ZDICT_params_t; 79 80/*! ZDICT_cover_params_t: 81 * k and d are the only required parameters. 82 * For others, value 0 means default. 83 */ 84typedef struct { 85 unsigned k; /* Segment size : constraint: 0 < k : Reasonable range [16, 2048+] */ 86 unsigned d; /* dmer size : constraint: 0 < d <= k : Reasonable range [6, 16] */ 87 unsigned steps; /* Number of steps : Only used for optimization : 0 means default (32) : Higher means more parameters checked */ 88 unsigned nbThreads; /* Number of threads : constraint: 0 < nbThreads : 1 means single-threaded : Only used for optimization : Ignored if ZSTD_MULTITHREAD is not defined */ 89 ZDICT_params_t zParams; 90} ZDICT_cover_params_t; 91 92 93/*! ZDICT_trainFromBuffer_cover(): 94 * Train a dictionary from an array of samples using the COVER algorithm. 95 * Samples must be stored concatenated in a single flat buffer `samplesBuffer`, 96 * supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order. 97 * The resulting dictionary will be saved into `dictBuffer`. 98 * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`) 99 * or an error code, which can be tested with ZDICT_isError(). 100 * Note: ZDICT_trainFromBuffer_cover() requires about 9 bytes of memory for each input byte. 101 * Tips: In general, a reasonable dictionary has a size of ~ 100 KB. 102 * It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`. 103 * In general, it's recommended to provide a few thousands samples, though this can vary a lot. 104 * It's recommended that total size of all samples be about ~x100 times the target size of dictionary. 105 */ 106ZDICTLIB_API size_t ZDICT_trainFromBuffer_cover( 107 void *dictBuffer, size_t dictBufferCapacity, 108 const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples, 109 ZDICT_cover_params_t parameters); 110 111/*! ZDICT_optimizeTrainFromBuffer_cover(): 112 * The same requirements as above hold for all the parameters except `parameters`. 113 * This function tries many parameter combinations and picks the best parameters. 114 * `*parameters` is filled with the best parameters found, 115 * dictionary constructed with those parameters is stored in `dictBuffer`. 116 * 117 * All of the parameters d, k, steps are optional. 118 * If d is non-zero then we don't check multiple values of d, otherwise we check d = {6, 8, 10, 12, 14, 16}. 119 * if steps is zero it defaults to its default value. 120 * If k is non-zero then we don't check multiple values of k, otherwise we check steps values in [16, 2048]. 121 * 122 * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`) 123 * or an error code, which can be tested with ZDICT_isError(). 124 * On success `*parameters` contains the parameters selected. 125 * 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. 126 */ 127ZDICTLIB_API size_t ZDICT_optimizeTrainFromBuffer_cover( 128 void* dictBuffer, size_t dictBufferCapacity, 129 const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples, 130 ZDICT_cover_params_t* parameters); 131 132/*! ZDICT_finalizeDictionary(): 133 * Given a custom content as a basis for dictionary, and a set of samples, 134 * finalize dictionary by adding headers and statistics. 135 * 136 * Samples must be stored concatenated in a flat buffer `samplesBuffer`, 137 * supplied with an array of sizes `samplesSizes`, providing the size of each sample in order. 138 * 139 * dictContentSize must be >= ZDICT_CONTENTSIZE_MIN bytes. 140 * maxDictSize must be >= dictContentSize, and must be >= ZDICT_DICTSIZE_MIN bytes. 141 * 142 * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`), 143 * or an error code, which can be tested by ZDICT_isError(). 144 * Note: ZDICT_finalizeDictionary() will push notifications into stderr if instructed to, using notificationLevel>0. 145 * Note 2: dictBuffer and dictContent can overlap 146 */ 147#define ZDICT_CONTENTSIZE_MIN 128 148#define ZDICT_DICTSIZE_MIN 256 149ZDICTLIB_API size_t ZDICT_finalizeDictionary(void* dictBuffer, size_t dictBufferCapacity, 150 const void* dictContent, size_t dictContentSize, 151 const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples, 152 ZDICT_params_t parameters); 153 154typedef struct { 155 unsigned selectivityLevel; /* 0 means default; larger => select more => larger dictionary */ 156 ZDICT_params_t zParams; 157} ZDICT_legacy_params_t; 158 159/*! ZDICT_trainFromBuffer_legacy(): 160 * Train a dictionary from an array of samples. 161 * Samples must be stored concatenated in a single flat buffer `samplesBuffer`, 162 * supplied with an array of sizes `samplesSizes`, providing the size of each sample, in order. 163 * The resulting dictionary will be saved into `dictBuffer`. 164 * `parameters` is optional and can be provided with values set to 0 to mean "default". 165 * @return: size of dictionary stored into `dictBuffer` (<= `dictBufferCapacity`) 166 * or an error code, which can be tested with ZDICT_isError(). 167 * Tips: In general, a reasonable dictionary has a size of ~ 100 KB. 168 * It's possible to select smaller or larger size, just by specifying `dictBufferCapacity`. 169 * In general, it's recommended to provide a few thousands samples, though this can vary a lot. 170 * It's recommended that total size of all samples be about ~x100 times the target size of dictionary. 171 * Note: ZDICT_trainFromBuffer_legacy() will send notifications into stderr if instructed to, using notificationLevel>0. 172 */ 173ZDICTLIB_API size_t ZDICT_trainFromBuffer_legacy( 174 void *dictBuffer, size_t dictBufferCapacity, 175 const void *samplesBuffer, const size_t *samplesSizes, unsigned nbSamples, 176 ZDICT_legacy_params_t parameters); 177 178/* Deprecation warnings */ 179/* It is generally possible to disable deprecation warnings from compiler, 180 for example with -Wno-deprecated-declarations for gcc 181 or _CRT_SECURE_NO_WARNINGS in Visual. 182 Otherwise, it's also possible to manually define ZDICT_DISABLE_DEPRECATE_WARNINGS */ 183#ifdef ZDICT_DISABLE_DEPRECATE_WARNINGS 184# define ZDICT_DEPRECATED(message) ZDICTLIB_API /* disable deprecation warnings */ 185#else 186# define ZDICT_GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) 187# if defined (__cplusplus) && (__cplusplus >= 201402) /* C++14 or greater */ 188# define ZDICT_DEPRECATED(message) [[deprecated(message)]] ZDICTLIB_API 189# elif (ZDICT_GCC_VERSION >= 405) || defined(__clang__) 190# define ZDICT_DEPRECATED(message) ZDICTLIB_API __attribute__((deprecated(message))) 191# elif (ZDICT_GCC_VERSION >= 301) 192# define ZDICT_DEPRECATED(message) ZDICTLIB_API __attribute__((deprecated)) 193# elif defined(_MSC_VER) 194# define ZDICT_DEPRECATED(message) ZDICTLIB_API __declspec(deprecated(message)) 195# else 196# pragma message("WARNING: You need to implement ZDICT_DEPRECATED for this compiler") 197# define ZDICT_DEPRECATED(message) ZDICTLIB_API 198# endif 199#endif /* ZDICT_DISABLE_DEPRECATE_WARNINGS */ 200 201ZDICT_DEPRECATED("use ZDICT_finalizeDictionary() instead") 202size_t ZDICT_addEntropyTablesFromBuffer(void* dictBuffer, size_t dictContentSize, size_t dictBufferCapacity, 203 const void* samplesBuffer, const size_t* samplesSizes, unsigned nbSamples); 204 205 206#endif /* ZDICT_STATIC_LINKING_ONLY */ 207 208#if defined (__cplusplus) 209} 210#endif 211 212#endif /* DICTBUILDER_H_001 */ 213