/src/duckdb/third_party/zstd/compress/zstd_compress.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * Copyright (c) 2016-2020, 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 | | /*-************************************* |
12 | | * Dependencies |
13 | | ***************************************/ |
14 | | #include <limits.h> /* INT_MAX */ |
15 | | #include <string.h> /* memset */ |
16 | | #include "zstd/common/mem.h" |
17 | | #include "zstd/compress/hist.h" /* HIST_countFast_wksp */ |
18 | | #include "zstd/common/fse.h" |
19 | | #include "zstd/common/fse_static.h" |
20 | | #include "zstd/common/huf.h" |
21 | | #include "zstd/common/huf_static.h" |
22 | | #include "zstd/compress/zstd_compress_internal.h" |
23 | | #include "zstd/compress/zstd_compress_sequences.h" |
24 | | #include "zstd/compress/zstd_compress_literals.h" |
25 | | #include "zstd/compress/zstd_fast.h" |
26 | | #include "zstd/compress/zstd_double_fast.h" |
27 | | #include "zstd/compress/zstd_lazy.h" |
28 | | #include "zstd/compress/zstd_opt.h" |
29 | | #include "zstd/compress/zstd_ldm.h" |
30 | | #include "zstd/compress/zstd_compress_superblock.h" |
31 | | |
32 | | |
33 | | namespace duckdb_zstd { |
34 | | /*-************************************* |
35 | | * Helper functions |
36 | | ***************************************/ |
37 | | /* ZSTD_compressBound() |
38 | | * Note that the result from this function is only compatible with the "normal" |
39 | | * full-block strategy. |
40 | | * When there are a lot of small blocks due to frequent flush in streaming mode |
41 | | * the overhead of headers can make the compressed data to be larger than the |
42 | | * return value of ZSTD_compressBound(). |
43 | | */ |
44 | 0 | size_t ZSTD_compressBound(size_t srcSize) { |
45 | 0 | return ZSTD_COMPRESSBOUND(srcSize); |
46 | 0 | } |
47 | | |
48 | | |
49 | | /*-************************************* |
50 | | * Context memory management |
51 | | ***************************************/ |
52 | | struct ZSTD_CDict_s { |
53 | | const void* dictContent; |
54 | | size_t dictContentSize; |
55 | | U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */ |
56 | | ZSTD_cwksp workspace; |
57 | | ZSTD_matchState_t matchState; |
58 | | ZSTD_compressedBlockState_t cBlockState; |
59 | | ZSTD_customMem customMem; |
60 | | U32 dictID; |
61 | | int compressionLevel; /* 0 indicates that advanced API was used to select CDict params */ |
62 | | }; /* typedef'd to ZSTD_CDict within "zstd.h" */ |
63 | | |
64 | | ZSTD_CCtx* ZSTD_createCCtx(void) |
65 | 0 | { |
66 | 0 | return ZSTD_createCCtx_advanced({NULL, NULL, NULL}); |
67 | 0 | } |
68 | | |
69 | | static void ZSTD_initCCtx(ZSTD_CCtx* cctx, ZSTD_customMem memManager) |
70 | 0 | { |
71 | 0 | assert(cctx != NULL); |
72 | 0 | memset(cctx, 0, sizeof(*cctx)); |
73 | 0 | cctx->customMem = memManager; |
74 | 0 | cctx->bmi2 = 0; |
75 | 0 | { size_t const err = ZSTD_CCtx_reset(cctx, ZSTD_reset_parameters); |
76 | 0 | assert(!ZSTD_isError(err)); |
77 | 0 | (void)err; |
78 | 0 | } |
79 | 0 | } |
80 | | |
81 | | ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem) |
82 | 0 | { |
83 | 0 | ZSTD_STATIC_ASSERT(zcss_init==0); |
84 | 0 | ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN==(0ULL - 1)); |
85 | 0 | if (!customMem.customAlloc ^ !customMem.customFree) return NULL; |
86 | 0 | { ZSTD_CCtx* const cctx = (ZSTD_CCtx*)ZSTD_malloc(sizeof(ZSTD_CCtx), customMem); |
87 | 0 | if (!cctx) return NULL; |
88 | 0 | ZSTD_initCCtx(cctx, customMem); |
89 | 0 | return cctx; |
90 | 0 | } |
91 | 0 | } |
92 | | |
93 | | ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize) |
94 | 0 | { |
95 | 0 | ZSTD_cwksp ws; |
96 | 0 | ZSTD_CCtx* cctx; |
97 | 0 | if (workspaceSize <= sizeof(ZSTD_CCtx)) return NULL; /* minimum size */ |
98 | 0 | if ((size_t)workspace & 7) return NULL; /* must be 8-aligned */ |
99 | 0 | ZSTD_cwksp_init(&ws, workspace, workspaceSize); |
100 | |
|
101 | 0 | cctx = (ZSTD_CCtx*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CCtx)); |
102 | 0 | if (cctx == NULL) return NULL; |
103 | | |
104 | 0 | memset(cctx, 0, sizeof(ZSTD_CCtx)); |
105 | 0 | ZSTD_cwksp_move(&cctx->workspace, &ws); |
106 | 0 | cctx->staticSize = workspaceSize; |
107 | | |
108 | | /* statically sized space. entropyWorkspace never moves (but prev/next block swap places) */ |
109 | 0 | if (!ZSTD_cwksp_check_available(&cctx->workspace, HUF_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t))) return NULL; |
110 | 0 | cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t)); |
111 | 0 | cctx->blockState.nextCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t)); |
112 | 0 | cctx->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cctx->workspace, HUF_WORKSPACE_SIZE); |
113 | 0 | cctx->bmi2 = 0; |
114 | 0 | return cctx; |
115 | 0 | } |
116 | | |
117 | | /** |
118 | | * Clears and frees all of the dictionaries in the CCtx. |
119 | | */ |
120 | | static void ZSTD_clearAllDicts(ZSTD_CCtx* cctx) |
121 | 0 | { |
122 | 0 | ZSTD_free(cctx->localDict.dictBuffer, cctx->customMem); |
123 | 0 | ZSTD_freeCDict(cctx->localDict.cdict); |
124 | 0 | memset(&cctx->localDict, 0, sizeof(cctx->localDict)); |
125 | 0 | memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict)); |
126 | 0 | cctx->cdict = NULL; |
127 | 0 | } |
128 | | |
129 | | static size_t ZSTD_sizeof_localDict(ZSTD_localDict dict) |
130 | 0 | { |
131 | 0 | size_t const bufferSize = dict.dictBuffer != NULL ? dict.dictSize : 0; |
132 | 0 | size_t const cdictSize = ZSTD_sizeof_CDict(dict.cdict); |
133 | 0 | return bufferSize + cdictSize; |
134 | 0 | } |
135 | | |
136 | | static void ZSTD_freeCCtxContent(ZSTD_CCtx* cctx) |
137 | 0 | { |
138 | 0 | assert(cctx != NULL); |
139 | 0 | assert(cctx->staticSize == 0); |
140 | 0 | ZSTD_clearAllDicts(cctx); |
141 | | #ifdef ZSTD_MULTITHREAD |
142 | | ZSTDMT_freeCCtx(cctx->mtctx); cctx->mtctx = NULL; |
143 | | #endif |
144 | 0 | ZSTD_cwksp_free(&cctx->workspace, cctx->customMem); |
145 | 0 | } |
146 | | |
147 | | size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx) |
148 | 0 | { |
149 | 0 | if (cctx==NULL) return 0; /* support free on NULL */ |
150 | 0 | RETURN_ERROR_IF(cctx->staticSize, memory_allocation, |
151 | 0 | "not compatible with static CCtx"); |
152 | 0 | { |
153 | 0 | int cctxInWorkspace = ZSTD_cwksp_owns_buffer(&cctx->workspace, cctx); |
154 | 0 | ZSTD_freeCCtxContent(cctx); |
155 | 0 | if (!cctxInWorkspace) { |
156 | 0 | ZSTD_free(cctx, cctx->customMem); |
157 | 0 | } |
158 | 0 | } |
159 | 0 | return 0; |
160 | 0 | } |
161 | | |
162 | | |
163 | | static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx) |
164 | 0 | { |
165 | | #ifdef ZSTD_MULTITHREAD |
166 | | return ZSTDMT_sizeof_CCtx(cctx->mtctx); |
167 | | #else |
168 | 0 | (void)cctx; |
169 | 0 | return 0; |
170 | 0 | #endif |
171 | 0 | } |
172 | | |
173 | | |
174 | | size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx) |
175 | 0 | { |
176 | 0 | if (cctx==NULL) return 0; /* support sizeof on NULL */ |
177 | | /* cctx may be in the workspace */ |
178 | 0 | return (cctx->workspace.workspace == cctx ? 0 : sizeof(*cctx)) |
179 | 0 | + ZSTD_cwksp_sizeof(&cctx->workspace) |
180 | 0 | + ZSTD_sizeof_localDict(cctx->localDict) |
181 | 0 | + ZSTD_sizeof_mtctx(cctx); |
182 | 0 | } |
183 | | |
184 | | size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs) |
185 | 0 | { |
186 | 0 | return ZSTD_sizeof_CCtx(zcs); /* same object */ |
187 | 0 | } |
188 | | |
189 | | /* private API call, for dictBuilder only */ |
190 | 0 | const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx) { return &(ctx->seqStore); } |
191 | | |
192 | | static ZSTD_CCtx_params ZSTD_makeCCtxParamsFromCParams( |
193 | | ZSTD_compressionParameters cParams) |
194 | 0 | { |
195 | 0 | ZSTD_CCtx_params cctxParams; |
196 | 0 | memset(&cctxParams, 0, sizeof(cctxParams)); |
197 | 0 | cctxParams.cParams = cParams; |
198 | 0 | cctxParams.compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */ |
199 | 0 | assert(!ZSTD_checkCParams(cParams)); |
200 | 0 | cctxParams.fParams.contentSizeFlag = 1; |
201 | 0 | return cctxParams; |
202 | 0 | } |
203 | | |
204 | | static ZSTD_CCtx_params* ZSTD_createCCtxParams_advanced( |
205 | | ZSTD_customMem customMem) |
206 | 0 | { |
207 | 0 | ZSTD_CCtx_params* params; |
208 | 0 | if (!customMem.customAlloc ^ !customMem.customFree) return NULL; |
209 | 0 | params = (ZSTD_CCtx_params*)ZSTD_calloc( |
210 | 0 | sizeof(ZSTD_CCtx_params), customMem); |
211 | 0 | if (!params) { return NULL; } |
212 | 0 | params->customMem = customMem; |
213 | 0 | params->compressionLevel = ZSTD_CLEVEL_DEFAULT; |
214 | 0 | params->fParams.contentSizeFlag = 1; |
215 | 0 | return params; |
216 | 0 | } |
217 | | |
218 | | ZSTD_CCtx_params* ZSTD_createCCtxParams(void) |
219 | 0 | { |
220 | 0 | return ZSTD_createCCtxParams_advanced(ZSTDInternalConstants::ZSTD_defaultCMem); |
221 | 0 | } |
222 | | |
223 | | size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params) |
224 | 0 | { |
225 | 0 | if (params == NULL) { return 0; } |
226 | 0 | ZSTD_free(params, params->customMem); |
227 | 0 | return 0; |
228 | 0 | } |
229 | | |
230 | | size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params) |
231 | 0 | { |
232 | 0 | return ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT); |
233 | 0 | } |
234 | | |
235 | 0 | size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) { |
236 | 0 | RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!"); |
237 | 0 | memset(cctxParams, 0, sizeof(*cctxParams)); |
238 | 0 | cctxParams->compressionLevel = compressionLevel; |
239 | 0 | cctxParams->fParams.contentSizeFlag = 1; |
240 | 0 | return 0; |
241 | 0 | } |
242 | | |
243 | | size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params) |
244 | 0 | { |
245 | 0 | RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!"); |
246 | 0 | FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , ""); |
247 | 0 | memset(cctxParams, 0, sizeof(*cctxParams)); |
248 | 0 | assert(!ZSTD_checkCParams(params.cParams)); |
249 | 0 | cctxParams->cParams = params.cParams; |
250 | 0 | cctxParams->fParams = params.fParams; |
251 | 0 | cctxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */ |
252 | 0 | return 0; |
253 | 0 | } |
254 | | |
255 | | /* ZSTD_assignParamsToCCtxParams() : |
256 | | * params is presumed valid at this stage */ |
257 | | static ZSTD_CCtx_params ZSTD_assignParamsToCCtxParams( |
258 | | const ZSTD_CCtx_params* cctxParams, const ZSTD_parameters* params) |
259 | 0 | { |
260 | 0 | ZSTD_CCtx_params ret = *cctxParams; |
261 | 0 | assert(!ZSTD_checkCParams(params->cParams)); |
262 | 0 | ret.cParams = params->cParams; |
263 | 0 | ret.fParams = params->fParams; |
264 | 0 | ret.compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */ |
265 | 0 | return ret; |
266 | 0 | } |
267 | | |
268 | | ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param) |
269 | 0 | { |
270 | 0 | ZSTD_bounds bounds = { 0, 0, 0 }; |
271 | |
|
272 | 0 | switch(param) |
273 | 0 | { |
274 | 0 | case ZSTD_c_compressionLevel: |
275 | 0 | bounds.lowerBound = ZSTD_minCLevel(); |
276 | 0 | bounds.upperBound = ZSTD_maxCLevel(); |
277 | 0 | return bounds; |
278 | | |
279 | 0 | case ZSTD_c_windowLog: |
280 | 0 | bounds.lowerBound = ZSTD_WINDOWLOG_MIN; |
281 | 0 | bounds.upperBound = ZSTD_WINDOWLOG_MAX; |
282 | 0 | return bounds; |
283 | | |
284 | 0 | case ZSTD_c_hashLog: |
285 | 0 | bounds.lowerBound = ZSTD_HASHLOG_MIN; |
286 | 0 | bounds.upperBound = ZSTD_HASHLOG_MAX; |
287 | 0 | return bounds; |
288 | | |
289 | 0 | case ZSTD_c_chainLog: |
290 | 0 | bounds.lowerBound = ZSTD_CHAINLOG_MIN; |
291 | 0 | bounds.upperBound = ZSTD_CHAINLOG_MAX; |
292 | 0 | return bounds; |
293 | | |
294 | 0 | case ZSTD_c_searchLog: |
295 | 0 | bounds.lowerBound = ZSTD_SEARCHLOG_MIN; |
296 | 0 | bounds.upperBound = ZSTD_SEARCHLOG_MAX; |
297 | 0 | return bounds; |
298 | | |
299 | 0 | case ZSTD_c_minMatch: |
300 | 0 | bounds.lowerBound = ZSTD_MINMATCH_MIN; |
301 | 0 | bounds.upperBound = ZSTD_MINMATCH_MAX; |
302 | 0 | return bounds; |
303 | | |
304 | 0 | case ZSTD_c_targetLength: |
305 | 0 | bounds.lowerBound = ZSTD_TARGETLENGTH_MIN; |
306 | 0 | bounds.upperBound = ZSTD_TARGETLENGTH_MAX; |
307 | 0 | return bounds; |
308 | | |
309 | 0 | case ZSTD_c_strategy: |
310 | 0 | bounds.lowerBound = ZSTD_STRATEGY_MIN; |
311 | 0 | bounds.upperBound = ZSTD_STRATEGY_MAX; |
312 | 0 | return bounds; |
313 | | |
314 | 0 | case ZSTD_c_contentSizeFlag: |
315 | 0 | bounds.lowerBound = 0; |
316 | 0 | bounds.upperBound = 1; |
317 | 0 | return bounds; |
318 | | |
319 | 0 | case ZSTD_c_checksumFlag: |
320 | 0 | bounds.lowerBound = 0; |
321 | 0 | bounds.upperBound = 1; |
322 | 0 | return bounds; |
323 | | |
324 | 0 | case ZSTD_c_dictIDFlag: |
325 | 0 | bounds.lowerBound = 0; |
326 | 0 | bounds.upperBound = 1; |
327 | 0 | return bounds; |
328 | | |
329 | 0 | case ZSTD_c_nbWorkers: |
330 | 0 | bounds.lowerBound = 0; |
331 | | #ifdef ZSTD_MULTITHREAD |
332 | | bounds.upperBound = ZSTDMT_NBWORKERS_MAX; |
333 | | #else |
334 | 0 | bounds.upperBound = 0; |
335 | 0 | #endif |
336 | 0 | return bounds; |
337 | | |
338 | 0 | case ZSTD_c_jobSize: |
339 | 0 | bounds.lowerBound = 0; |
340 | | #ifdef ZSTD_MULTITHREAD |
341 | | bounds.upperBound = ZSTDMT_JOBSIZE_MAX; |
342 | | #else |
343 | 0 | bounds.upperBound = 0; |
344 | 0 | #endif |
345 | 0 | return bounds; |
346 | | |
347 | 0 | case ZSTD_c_overlapLog: |
348 | | #ifdef ZSTD_MULTITHREAD |
349 | | bounds.lowerBound = ZSTD_OVERLAPLOG_MIN; |
350 | | bounds.upperBound = ZSTD_OVERLAPLOG_MAX; |
351 | | #else |
352 | 0 | bounds.lowerBound = 0; |
353 | 0 | bounds.upperBound = 0; |
354 | 0 | #endif |
355 | 0 | return bounds; |
356 | | |
357 | 0 | case ZSTD_c_enableLongDistanceMatching: |
358 | 0 | bounds.lowerBound = 0; |
359 | 0 | bounds.upperBound = 1; |
360 | 0 | return bounds; |
361 | | |
362 | 0 | case ZSTD_c_ldmHashLog: |
363 | 0 | bounds.lowerBound = ZSTD_LDM_HASHLOG_MIN; |
364 | 0 | bounds.upperBound = ZSTD_LDM_HASHLOG_MAX; |
365 | 0 | return bounds; |
366 | | |
367 | 0 | case ZSTD_c_ldmMinMatch: |
368 | 0 | bounds.lowerBound = ZSTD_LDM_MINMATCH_MIN; |
369 | 0 | bounds.upperBound = ZSTD_LDM_MINMATCH_MAX; |
370 | 0 | return bounds; |
371 | | |
372 | 0 | case ZSTD_c_ldmBucketSizeLog: |
373 | 0 | bounds.lowerBound = ZSTD_LDM_BUCKETSIZELOG_MIN; |
374 | 0 | bounds.upperBound = ZSTD_LDM_BUCKETSIZELOG_MAX; |
375 | 0 | return bounds; |
376 | | |
377 | 0 | case ZSTD_c_ldmHashRateLog: |
378 | 0 | bounds.lowerBound = ZSTD_LDM_HASHRATELOG_MIN; |
379 | 0 | bounds.upperBound = ZSTD_LDM_HASHRATELOG_MAX; |
380 | 0 | return bounds; |
381 | | |
382 | | /* experimental parameters */ |
383 | 0 | case ZSTD_c_rsyncable: |
384 | 0 | bounds.lowerBound = 0; |
385 | 0 | bounds.upperBound = 1; |
386 | 0 | return bounds; |
387 | | |
388 | 0 | case ZSTD_c_forceMaxWindow : |
389 | 0 | bounds.lowerBound = 0; |
390 | 0 | bounds.upperBound = 1; |
391 | 0 | return bounds; |
392 | | |
393 | 0 | case ZSTD_c_format: |
394 | 0 | ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless); |
395 | 0 | bounds.lowerBound = ZSTD_f_zstd1; |
396 | 0 | bounds.upperBound = ZSTD_f_zstd1_magicless; /* note : how to ensure at compile time that this is the highest value enum ? */ |
397 | 0 | return bounds; |
398 | | |
399 | 0 | case ZSTD_c_forceAttachDict: |
400 | 0 | ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceCopy); |
401 | 0 | bounds.lowerBound = ZSTD_dictDefaultAttach; |
402 | 0 | bounds.upperBound = ZSTD_dictForceLoad; /* note : how to ensure at compile time that this is the highest value enum ? */ |
403 | 0 | return bounds; |
404 | | |
405 | 0 | case ZSTD_c_literalCompressionMode: |
406 | 0 | ZSTD_STATIC_ASSERT(ZSTD_lcm_auto < ZSTD_lcm_huffman && ZSTD_lcm_huffman < ZSTD_lcm_uncompressed); |
407 | 0 | bounds.lowerBound = ZSTD_lcm_auto; |
408 | 0 | bounds.upperBound = ZSTD_lcm_uncompressed; |
409 | 0 | return bounds; |
410 | | |
411 | 0 | case ZSTD_c_targetCBlockSize: |
412 | 0 | bounds.lowerBound = ZSTD_TARGETCBLOCKSIZE_MIN; |
413 | 0 | bounds.upperBound = ZSTD_TARGETCBLOCKSIZE_MAX; |
414 | 0 | return bounds; |
415 | | |
416 | 0 | case ZSTD_c_srcSizeHint: |
417 | 0 | bounds.lowerBound = ZSTD_SRCSIZEHINT_MIN; |
418 | 0 | bounds.upperBound = ZSTD_SRCSIZEHINT_MAX; |
419 | 0 | return bounds; |
420 | | |
421 | 0 | default: |
422 | 0 | bounds.error = ERROR(parameter_unsupported); |
423 | 0 | return bounds; |
424 | 0 | } |
425 | 0 | } |
426 | | |
427 | | /* ZSTD_cParam_clampBounds: |
428 | | * Clamps the value into the bounded range. |
429 | | */ |
430 | | static size_t ZSTD_cParam_clampBounds(ZSTD_cParameter cParam, int* value) |
431 | 0 | { |
432 | 0 | ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam); |
433 | 0 | if (ZSTD_isError(bounds.error)) return bounds.error; |
434 | 0 | if (*value < bounds.lowerBound) *value = bounds.lowerBound; |
435 | 0 | if (*value > bounds.upperBound) *value = bounds.upperBound; |
436 | 0 | return 0; |
437 | 0 | } |
438 | | |
439 | 0 | #define BOUNDCHECK(cParam, val) { \ |
440 | 0 | RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \ |
441 | 0 | parameter_outOfBound, "Param out of bounds"); \ |
442 | 0 | } |
443 | | |
444 | | |
445 | | static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param) |
446 | 0 | { |
447 | 0 | switch(param) |
448 | 0 | { |
449 | 0 | case ZSTD_c_compressionLevel: |
450 | 0 | case ZSTD_c_hashLog: |
451 | 0 | case ZSTD_c_chainLog: |
452 | 0 | case ZSTD_c_searchLog: |
453 | 0 | case ZSTD_c_minMatch: |
454 | 0 | case ZSTD_c_targetLength: |
455 | 0 | case ZSTD_c_strategy: |
456 | 0 | return 1; |
457 | | |
458 | 0 | case ZSTD_c_format: |
459 | 0 | case ZSTD_c_windowLog: |
460 | 0 | case ZSTD_c_contentSizeFlag: |
461 | 0 | case ZSTD_c_checksumFlag: |
462 | 0 | case ZSTD_c_dictIDFlag: |
463 | 0 | case ZSTD_c_forceMaxWindow : |
464 | 0 | case ZSTD_c_nbWorkers: |
465 | 0 | case ZSTD_c_jobSize: |
466 | 0 | case ZSTD_c_overlapLog: |
467 | 0 | case ZSTD_c_rsyncable: |
468 | 0 | case ZSTD_c_enableLongDistanceMatching: |
469 | 0 | case ZSTD_c_ldmHashLog: |
470 | 0 | case ZSTD_c_ldmMinMatch: |
471 | 0 | case ZSTD_c_ldmBucketSizeLog: |
472 | 0 | case ZSTD_c_ldmHashRateLog: |
473 | 0 | case ZSTD_c_forceAttachDict: |
474 | 0 | case ZSTD_c_literalCompressionMode: |
475 | 0 | case ZSTD_c_targetCBlockSize: |
476 | 0 | case ZSTD_c_srcSizeHint: |
477 | 0 | default: |
478 | 0 | return 0; |
479 | 0 | } |
480 | 0 | } |
481 | | |
482 | | size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value) |
483 | 0 | { |
484 | 0 | DEBUGLOG(4, "ZSTD_CCtx_setParameter (%i, %i)", (int)param, value); |
485 | 0 | if (cctx->streamStage != zcss_init) { |
486 | 0 | if (ZSTD_isUpdateAuthorized(param)) { |
487 | 0 | cctx->cParamsChanged = 1; |
488 | 0 | } else { |
489 | 0 | RETURN_ERROR(stage_wrong, "can only set params in ctx init stage"); |
490 | 0 | } } |
491 | | |
492 | 0 | switch(param) |
493 | 0 | { |
494 | 0 | case ZSTD_c_nbWorkers: |
495 | 0 | RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported, |
496 | 0 | "MT not compatible with static alloc"); |
497 | 0 | break; |
498 | | |
499 | 0 | case ZSTD_c_compressionLevel: |
500 | 0 | case ZSTD_c_windowLog: |
501 | 0 | case ZSTD_c_hashLog: |
502 | 0 | case ZSTD_c_chainLog: |
503 | 0 | case ZSTD_c_searchLog: |
504 | 0 | case ZSTD_c_minMatch: |
505 | 0 | case ZSTD_c_targetLength: |
506 | 0 | case ZSTD_c_strategy: |
507 | 0 | case ZSTD_c_ldmHashRateLog: |
508 | 0 | case ZSTD_c_format: |
509 | 0 | case ZSTD_c_contentSizeFlag: |
510 | 0 | case ZSTD_c_checksumFlag: |
511 | 0 | case ZSTD_c_dictIDFlag: |
512 | 0 | case ZSTD_c_forceMaxWindow: |
513 | 0 | case ZSTD_c_forceAttachDict: |
514 | 0 | case ZSTD_c_literalCompressionMode: |
515 | 0 | case ZSTD_c_jobSize: |
516 | 0 | case ZSTD_c_overlapLog: |
517 | 0 | case ZSTD_c_rsyncable: |
518 | 0 | case ZSTD_c_enableLongDistanceMatching: |
519 | 0 | case ZSTD_c_ldmHashLog: |
520 | 0 | case ZSTD_c_ldmMinMatch: |
521 | 0 | case ZSTD_c_ldmBucketSizeLog: |
522 | 0 | case ZSTD_c_targetCBlockSize: |
523 | 0 | case ZSTD_c_srcSizeHint: |
524 | 0 | break; |
525 | | |
526 | 0 | default: RETURN_ERROR(parameter_unsupported, "unknown parameter"); |
527 | 0 | } |
528 | 0 | return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value); |
529 | 0 | } |
530 | | |
531 | | size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams, |
532 | | ZSTD_cParameter param, int value) |
533 | 0 | { |
534 | 0 | DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value); |
535 | 0 | switch(param) |
536 | 0 | { |
537 | 0 | case ZSTD_c_format : |
538 | 0 | BOUNDCHECK(ZSTD_c_format, value); |
539 | 0 | CCtxParams->format = (ZSTD_format_e)value; |
540 | 0 | return (size_t)CCtxParams->format; |
541 | | |
542 | 0 | case ZSTD_c_compressionLevel : { |
543 | 0 | FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), ""); |
544 | 0 | if (value) { /* 0 : does not change current level */ |
545 | 0 | CCtxParams->compressionLevel = value; |
546 | 0 | } |
547 | 0 | if (CCtxParams->compressionLevel >= 0) return (size_t)CCtxParams->compressionLevel; |
548 | 0 | return 0; /* return type (size_t) cannot represent negative values */ |
549 | 0 | } |
550 | | |
551 | 0 | case ZSTD_c_windowLog : |
552 | 0 | if (value!=0) /* 0 => use default */ |
553 | 0 | BOUNDCHECK(ZSTD_c_windowLog, value); |
554 | 0 | CCtxParams->cParams.windowLog = (U32)value; |
555 | 0 | return CCtxParams->cParams.windowLog; |
556 | | |
557 | 0 | case ZSTD_c_hashLog : |
558 | 0 | if (value!=0) /* 0 => use default */ |
559 | 0 | BOUNDCHECK(ZSTD_c_hashLog, value); |
560 | 0 | CCtxParams->cParams.hashLog = (U32)value; |
561 | 0 | return CCtxParams->cParams.hashLog; |
562 | | |
563 | 0 | case ZSTD_c_chainLog : |
564 | 0 | if (value!=0) /* 0 => use default */ |
565 | 0 | BOUNDCHECK(ZSTD_c_chainLog, value); |
566 | 0 | CCtxParams->cParams.chainLog = (U32)value; |
567 | 0 | return CCtxParams->cParams.chainLog; |
568 | | |
569 | 0 | case ZSTD_c_searchLog : |
570 | 0 | if (value!=0) /* 0 => use default */ |
571 | 0 | BOUNDCHECK(ZSTD_c_searchLog, value); |
572 | 0 | CCtxParams->cParams.searchLog = (U32)value; |
573 | 0 | return (size_t)value; |
574 | | |
575 | 0 | case ZSTD_c_minMatch : |
576 | 0 | if (value!=0) /* 0 => use default */ |
577 | 0 | BOUNDCHECK(ZSTD_c_minMatch, value); |
578 | 0 | CCtxParams->cParams.minMatch = value; |
579 | 0 | return CCtxParams->cParams.minMatch; |
580 | | |
581 | 0 | case ZSTD_c_targetLength : |
582 | 0 | BOUNDCHECK(ZSTD_c_targetLength, value); |
583 | 0 | CCtxParams->cParams.targetLength = value; |
584 | 0 | return CCtxParams->cParams.targetLength; |
585 | | |
586 | 0 | case ZSTD_c_strategy : |
587 | 0 | if (value!=0) /* 0 => use default */ |
588 | 0 | BOUNDCHECK(ZSTD_c_strategy, value); |
589 | 0 | CCtxParams->cParams.strategy = (ZSTD_strategy)value; |
590 | 0 | return (size_t)CCtxParams->cParams.strategy; |
591 | | |
592 | 0 | case ZSTD_c_contentSizeFlag : |
593 | | /* Content size written in frame header _when known_ (default:1) */ |
594 | 0 | DEBUGLOG(4, "set content size flag = %u", (value!=0)); |
595 | 0 | CCtxParams->fParams.contentSizeFlag = value != 0; |
596 | 0 | return CCtxParams->fParams.contentSizeFlag; |
597 | | |
598 | 0 | case ZSTD_c_checksumFlag : |
599 | | /* A 32-bits content checksum will be calculated and written at end of frame (default:0) */ |
600 | 0 | CCtxParams->fParams.checksumFlag = value != 0; |
601 | 0 | return CCtxParams->fParams.checksumFlag; |
602 | | |
603 | 0 | case ZSTD_c_dictIDFlag : /* When applicable, dictionary's dictID is provided in frame header (default:1) */ |
604 | 0 | DEBUGLOG(4, "set dictIDFlag = %u", (value!=0)); |
605 | 0 | CCtxParams->fParams.noDictIDFlag = !value; |
606 | 0 | return !CCtxParams->fParams.noDictIDFlag; |
607 | | |
608 | 0 | case ZSTD_c_forceMaxWindow : |
609 | 0 | CCtxParams->forceWindow = (value != 0); |
610 | 0 | return CCtxParams->forceWindow; |
611 | | |
612 | 0 | case ZSTD_c_forceAttachDict : { |
613 | 0 | const ZSTD_dictAttachPref_e pref = (ZSTD_dictAttachPref_e)value; |
614 | 0 | BOUNDCHECK(ZSTD_c_forceAttachDict, pref); |
615 | 0 | CCtxParams->attachDictPref = pref; |
616 | 0 | return CCtxParams->attachDictPref; |
617 | 0 | } |
618 | | |
619 | 0 | case ZSTD_c_literalCompressionMode : { |
620 | 0 | const ZSTD_literalCompressionMode_e lcm = (ZSTD_literalCompressionMode_e)value; |
621 | 0 | BOUNDCHECK(ZSTD_c_literalCompressionMode, lcm); |
622 | 0 | CCtxParams->literalCompressionMode = lcm; |
623 | 0 | return CCtxParams->literalCompressionMode; |
624 | 0 | } |
625 | | |
626 | 0 | case ZSTD_c_nbWorkers : |
627 | 0 | #ifndef ZSTD_MULTITHREAD |
628 | 0 | RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading"); |
629 | 0 | return 0; |
630 | | #else |
631 | | FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), ""); |
632 | | CCtxParams->nbWorkers = value; |
633 | | return CCtxParams->nbWorkers; |
634 | | #endif |
635 | | |
636 | 0 | case ZSTD_c_jobSize : |
637 | 0 | #ifndef ZSTD_MULTITHREAD |
638 | 0 | RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading"); |
639 | 0 | return 0; |
640 | | #else |
641 | | /* Adjust to the minimum non-default value. */ |
642 | | if (value != 0 && value < ZSTDMT_JOBSIZE_MIN) |
643 | | value = ZSTDMT_JOBSIZE_MIN; |
644 | | FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), ""); |
645 | | assert(value >= 0); |
646 | | CCtxParams->jobSize = value; |
647 | | return CCtxParams->jobSize; |
648 | | #endif |
649 | | |
650 | 0 | case ZSTD_c_overlapLog : |
651 | 0 | #ifndef ZSTD_MULTITHREAD |
652 | 0 | RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading"); |
653 | 0 | return 0; |
654 | | #else |
655 | | FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), ""); |
656 | | CCtxParams->overlapLog = value; |
657 | | return CCtxParams->overlapLog; |
658 | | #endif |
659 | | |
660 | 0 | case ZSTD_c_rsyncable : |
661 | 0 | #ifndef ZSTD_MULTITHREAD |
662 | 0 | RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading"); |
663 | 0 | return 0; |
664 | | #else |
665 | | FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), ""); |
666 | | CCtxParams->rsyncable = value; |
667 | | return CCtxParams->rsyncable; |
668 | | #endif |
669 | | |
670 | 0 | case ZSTD_c_enableLongDistanceMatching : |
671 | 0 | CCtxParams->ldmParams.enableLdm = (value!=0); |
672 | 0 | return CCtxParams->ldmParams.enableLdm; |
673 | | |
674 | 0 | case ZSTD_c_ldmHashLog : |
675 | 0 | if (value!=0) /* 0 ==> auto */ |
676 | 0 | BOUNDCHECK(ZSTD_c_ldmHashLog, value); |
677 | 0 | CCtxParams->ldmParams.hashLog = value; |
678 | 0 | return CCtxParams->ldmParams.hashLog; |
679 | | |
680 | 0 | case ZSTD_c_ldmMinMatch : |
681 | 0 | if (value!=0) /* 0 ==> default */ |
682 | 0 | BOUNDCHECK(ZSTD_c_ldmMinMatch, value); |
683 | 0 | CCtxParams->ldmParams.minMatchLength = value; |
684 | 0 | return CCtxParams->ldmParams.minMatchLength; |
685 | | |
686 | 0 | case ZSTD_c_ldmBucketSizeLog : |
687 | 0 | if (value!=0) /* 0 ==> default */ |
688 | 0 | BOUNDCHECK(ZSTD_c_ldmBucketSizeLog, value); |
689 | 0 | CCtxParams->ldmParams.bucketSizeLog = value; |
690 | 0 | return CCtxParams->ldmParams.bucketSizeLog; |
691 | | |
692 | 0 | case ZSTD_c_ldmHashRateLog : |
693 | 0 | RETURN_ERROR_IF(value > ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN, |
694 | 0 | parameter_outOfBound, "Param out of bounds!"); |
695 | 0 | CCtxParams->ldmParams.hashRateLog = value; |
696 | 0 | return CCtxParams->ldmParams.hashRateLog; |
697 | | |
698 | 0 | case ZSTD_c_targetCBlockSize : |
699 | 0 | if (value!=0) /* 0 ==> default */ |
700 | 0 | BOUNDCHECK(ZSTD_c_targetCBlockSize, value); |
701 | 0 | CCtxParams->targetCBlockSize = value; |
702 | 0 | return CCtxParams->targetCBlockSize; |
703 | | |
704 | 0 | case ZSTD_c_srcSizeHint : |
705 | 0 | if (value!=0) /* 0 ==> default */ |
706 | 0 | BOUNDCHECK(ZSTD_c_srcSizeHint, value); |
707 | 0 | CCtxParams->srcSizeHint = value; |
708 | 0 | return CCtxParams->srcSizeHint; |
709 | | |
710 | 0 | default: RETURN_ERROR(parameter_unsupported, "unknown parameter"); |
711 | 0 | } |
712 | 0 | } |
713 | | |
714 | | size_t ZSTD_CCtx_getParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value) |
715 | 0 | { |
716 | 0 | return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value); |
717 | 0 | } |
718 | | |
719 | | size_t ZSTD_CCtxParams_getParameter( |
720 | | ZSTD_CCtx_params* CCtxParams, ZSTD_cParameter param, int* value) |
721 | 0 | { |
722 | 0 | switch(param) |
723 | 0 | { |
724 | 0 | case ZSTD_c_format : |
725 | 0 | *value = CCtxParams->format; |
726 | 0 | break; |
727 | 0 | case ZSTD_c_compressionLevel : |
728 | 0 | *value = CCtxParams->compressionLevel; |
729 | 0 | break; |
730 | 0 | case ZSTD_c_windowLog : |
731 | 0 | *value = (int)CCtxParams->cParams.windowLog; |
732 | 0 | break; |
733 | 0 | case ZSTD_c_hashLog : |
734 | 0 | *value = (int)CCtxParams->cParams.hashLog; |
735 | 0 | break; |
736 | 0 | case ZSTD_c_chainLog : |
737 | 0 | *value = (int)CCtxParams->cParams.chainLog; |
738 | 0 | break; |
739 | 0 | case ZSTD_c_searchLog : |
740 | 0 | *value = CCtxParams->cParams.searchLog; |
741 | 0 | break; |
742 | 0 | case ZSTD_c_minMatch : |
743 | 0 | *value = CCtxParams->cParams.minMatch; |
744 | 0 | break; |
745 | 0 | case ZSTD_c_targetLength : |
746 | 0 | *value = CCtxParams->cParams.targetLength; |
747 | 0 | break; |
748 | 0 | case ZSTD_c_strategy : |
749 | 0 | *value = (unsigned)CCtxParams->cParams.strategy; |
750 | 0 | break; |
751 | 0 | case ZSTD_c_contentSizeFlag : |
752 | 0 | *value = CCtxParams->fParams.contentSizeFlag; |
753 | 0 | break; |
754 | 0 | case ZSTD_c_checksumFlag : |
755 | 0 | *value = CCtxParams->fParams.checksumFlag; |
756 | 0 | break; |
757 | 0 | case ZSTD_c_dictIDFlag : |
758 | 0 | *value = !CCtxParams->fParams.noDictIDFlag; |
759 | 0 | break; |
760 | 0 | case ZSTD_c_forceMaxWindow : |
761 | 0 | *value = CCtxParams->forceWindow; |
762 | 0 | break; |
763 | 0 | case ZSTD_c_forceAttachDict : |
764 | 0 | *value = CCtxParams->attachDictPref; |
765 | 0 | break; |
766 | 0 | case ZSTD_c_literalCompressionMode : |
767 | 0 | *value = CCtxParams->literalCompressionMode; |
768 | 0 | break; |
769 | 0 | case ZSTD_c_nbWorkers : |
770 | 0 | #ifndef ZSTD_MULTITHREAD |
771 | 0 | assert(CCtxParams->nbWorkers == 0); |
772 | 0 | #endif |
773 | 0 | *value = CCtxParams->nbWorkers; |
774 | 0 | break; |
775 | 0 | case ZSTD_c_jobSize : |
776 | 0 | #ifndef ZSTD_MULTITHREAD |
777 | 0 | RETURN_ERROR(parameter_unsupported, "not compiled with multithreading"); |
778 | | #else |
779 | | assert(CCtxParams->jobSize <= INT_MAX); |
780 | | *value = (int)CCtxParams->jobSize; |
781 | | break; |
782 | | #endif |
783 | 0 | case ZSTD_c_overlapLog : |
784 | 0 | #ifndef ZSTD_MULTITHREAD |
785 | 0 | RETURN_ERROR(parameter_unsupported, "not compiled with multithreading"); |
786 | | #else |
787 | | *value = CCtxParams->overlapLog; |
788 | | break; |
789 | | #endif |
790 | 0 | case ZSTD_c_rsyncable : |
791 | 0 | #ifndef ZSTD_MULTITHREAD |
792 | 0 | RETURN_ERROR(parameter_unsupported, "not compiled with multithreading"); |
793 | | #else |
794 | | *value = CCtxParams->rsyncable; |
795 | | break; |
796 | | #endif |
797 | 0 | case ZSTD_c_enableLongDistanceMatching : |
798 | 0 | *value = CCtxParams->ldmParams.enableLdm; |
799 | 0 | break; |
800 | 0 | case ZSTD_c_ldmHashLog : |
801 | 0 | *value = CCtxParams->ldmParams.hashLog; |
802 | 0 | break; |
803 | 0 | case ZSTD_c_ldmMinMatch : |
804 | 0 | *value = CCtxParams->ldmParams.minMatchLength; |
805 | 0 | break; |
806 | 0 | case ZSTD_c_ldmBucketSizeLog : |
807 | 0 | *value = CCtxParams->ldmParams.bucketSizeLog; |
808 | 0 | break; |
809 | 0 | case ZSTD_c_ldmHashRateLog : |
810 | 0 | *value = CCtxParams->ldmParams.hashRateLog; |
811 | 0 | break; |
812 | 0 | case ZSTD_c_targetCBlockSize : |
813 | 0 | *value = (int)CCtxParams->targetCBlockSize; |
814 | 0 | break; |
815 | 0 | case ZSTD_c_srcSizeHint : |
816 | 0 | *value = (int)CCtxParams->srcSizeHint; |
817 | 0 | break; |
818 | 0 | default: RETURN_ERROR(parameter_unsupported, "unknown parameter"); |
819 | 0 | } |
820 | 0 | return 0; |
821 | 0 | } |
822 | | |
823 | | /** ZSTD_CCtx_setParametersUsingCCtxParams() : |
824 | | * just applies `params` into `cctx` |
825 | | * no action is performed, parameters are merely stored. |
826 | | * If ZSTDMT is enabled, parameters are pushed to cctx->mtctx. |
827 | | * This is possible even if a compression is ongoing. |
828 | | * In which case, new parameters will be applied on the fly, starting with next compression job. |
829 | | */ |
830 | | size_t ZSTD_CCtx_setParametersUsingCCtxParams( |
831 | | ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params) |
832 | 0 | { |
833 | 0 | DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams"); |
834 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
835 | 0 | "The context is in the wrong stage!"); |
836 | 0 | RETURN_ERROR_IF(cctx->cdict, stage_wrong, |
837 | 0 | "Can't override parameters with cdict attached (some must " |
838 | 0 | "be inherited from the cdict)."); |
839 | |
|
840 | 0 | cctx->requestedParams = *params; |
841 | 0 | return 0; |
842 | 0 | } |
843 | | |
844 | | ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize) |
845 | 0 | { |
846 | 0 | DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %u bytes", (U32)pledgedSrcSize); |
847 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
848 | 0 | "Can't set pledgedSrcSize when not in init stage."); |
849 | 0 | cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1; |
850 | 0 | return 0; |
851 | 0 | } |
852 | | |
853 | | /** |
854 | | * Initializes the local dict using the requested parameters. |
855 | | * NOTE: This does not use the pledged src size, because it may be used for more |
856 | | * than one compression. |
857 | | */ |
858 | | static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx) |
859 | 0 | { |
860 | 0 | ZSTD_localDict* const dl = &cctx->localDict; |
861 | 0 | ZSTD_compressionParameters const cParams = ZSTD_getCParamsFromCCtxParams( |
862 | 0 | &cctx->requestedParams, ZSTD_CONTENTSIZE_UNKNOWN, dl->dictSize); |
863 | 0 | if (dl->dict == NULL) { |
864 | | /* No local dictionary. */ |
865 | 0 | assert(dl->dictBuffer == NULL); |
866 | 0 | assert(dl->cdict == NULL); |
867 | 0 | assert(dl->dictSize == 0); |
868 | 0 | return 0; |
869 | 0 | } |
870 | 0 | if (dl->cdict != NULL) { |
871 | 0 | assert(cctx->cdict == dl->cdict); |
872 | | /* Local dictionary already initialized. */ |
873 | 0 | return 0; |
874 | 0 | } |
875 | 0 | assert(dl->dictSize > 0); |
876 | 0 | assert(cctx->cdict == NULL); |
877 | 0 | assert(cctx->prefixDict.dict == NULL); |
878 | |
|
879 | 0 | dl->cdict = ZSTD_createCDict_advanced( |
880 | 0 | dl->dict, |
881 | 0 | dl->dictSize, |
882 | 0 | ZSTD_dlm_byRef, |
883 | 0 | dl->dictContentType, |
884 | 0 | cParams, |
885 | 0 | cctx->customMem); |
886 | 0 | RETURN_ERROR_IF(!dl->cdict, memory_allocation, "ZSTD_createCDict_advanced failed"); |
887 | 0 | cctx->cdict = dl->cdict; |
888 | 0 | return 0; |
889 | 0 | } |
890 | | |
891 | | size_t ZSTD_CCtx_loadDictionary_advanced( |
892 | | ZSTD_CCtx* cctx, const void* dict, size_t dictSize, |
893 | | ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType) |
894 | 0 | { |
895 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
896 | 0 | "Can't load a dictionary when ctx is not in init stage."); |
897 | 0 | RETURN_ERROR_IF(cctx->staticSize, memory_allocation, |
898 | 0 | "no malloc for static CCtx"); |
899 | 0 | DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize); |
900 | 0 | ZSTD_clearAllDicts(cctx); /* in case one already exists */ |
901 | 0 | if (dict == NULL || dictSize == 0) /* no dictionary mode */ |
902 | 0 | return 0; |
903 | 0 | if (dictLoadMethod == ZSTD_dlm_byRef) { |
904 | 0 | cctx->localDict.dict = dict; |
905 | 0 | } else { |
906 | 0 | void* dictBuffer = ZSTD_malloc(dictSize, cctx->customMem); |
907 | 0 | RETURN_ERROR_IF(!dictBuffer, memory_allocation, "NULL pointer!"); |
908 | 0 | memcpy(dictBuffer, dict, dictSize); |
909 | 0 | cctx->localDict.dictBuffer = dictBuffer; |
910 | 0 | cctx->localDict.dict = dictBuffer; |
911 | 0 | } |
912 | 0 | cctx->localDict.dictSize = dictSize; |
913 | 0 | cctx->localDict.dictContentType = dictContentType; |
914 | 0 | return 0; |
915 | 0 | } |
916 | | |
917 | | ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference( |
918 | | ZSTD_CCtx* cctx, const void* dict, size_t dictSize) |
919 | 0 | { |
920 | 0 | return ZSTD_CCtx_loadDictionary_advanced( |
921 | 0 | cctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto); |
922 | 0 | } |
923 | | |
924 | | ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize) |
925 | 0 | { |
926 | 0 | return ZSTD_CCtx_loadDictionary_advanced( |
927 | 0 | cctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto); |
928 | 0 | } |
929 | | |
930 | | |
931 | | size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict) |
932 | 0 | { |
933 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
934 | 0 | "Can't ref a dict when ctx not in init stage."); |
935 | | /* Free the existing local cdict (if any) to save memory. */ |
936 | 0 | ZSTD_clearAllDicts(cctx); |
937 | 0 | cctx->cdict = cdict; |
938 | 0 | return 0; |
939 | 0 | } |
940 | | |
941 | | size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize) |
942 | 0 | { |
943 | 0 | return ZSTD_CCtx_refPrefix_advanced(cctx, prefix, prefixSize, ZSTD_dct_rawContent); |
944 | 0 | } |
945 | | |
946 | | size_t ZSTD_CCtx_refPrefix_advanced( |
947 | | ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType) |
948 | 0 | { |
949 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
950 | 0 | "Can't ref a prefix when ctx not in init stage."); |
951 | 0 | ZSTD_clearAllDicts(cctx); |
952 | 0 | if (prefix != NULL && prefixSize > 0) { |
953 | 0 | cctx->prefixDict.dict = prefix; |
954 | 0 | cctx->prefixDict.dictSize = prefixSize; |
955 | 0 | cctx->prefixDict.dictContentType = dictContentType; |
956 | 0 | } |
957 | 0 | return 0; |
958 | 0 | } |
959 | | |
960 | | /*! ZSTD_CCtx_reset() : |
961 | | * Also dumps dictionary */ |
962 | | size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset) |
963 | 0 | { |
964 | 0 | if ( (reset == ZSTD_reset_session_only) |
965 | 0 | || (reset == ZSTD_reset_session_and_parameters) ) { |
966 | 0 | cctx->streamStage = zcss_init; |
967 | 0 | cctx->pledgedSrcSizePlusOne = 0; |
968 | 0 | } |
969 | 0 | if ( (reset == ZSTD_reset_parameters) |
970 | 0 | || (reset == ZSTD_reset_session_and_parameters) ) { |
971 | 0 | RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong, |
972 | 0 | "Can't reset parameters only when not in init stage."); |
973 | 0 | ZSTD_clearAllDicts(cctx); |
974 | 0 | return ZSTD_CCtxParams_reset(&cctx->requestedParams); |
975 | 0 | } |
976 | 0 | return 0; |
977 | 0 | } |
978 | | |
979 | | |
980 | | /** ZSTD_checkCParams() : |
981 | | control CParam values remain within authorized range. |
982 | | @return : 0, or an error code if one value is beyond authorized range */ |
983 | | size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams) |
984 | 0 | { |
985 | 0 | BOUNDCHECK(ZSTD_c_windowLog, (int)cParams.windowLog); |
986 | 0 | BOUNDCHECK(ZSTD_c_chainLog, (int)cParams.chainLog); |
987 | 0 | BOUNDCHECK(ZSTD_c_hashLog, (int)cParams.hashLog); |
988 | 0 | BOUNDCHECK(ZSTD_c_searchLog, (int)cParams.searchLog); |
989 | 0 | BOUNDCHECK(ZSTD_c_minMatch, (int)cParams.minMatch); |
990 | 0 | BOUNDCHECK(ZSTD_c_targetLength,(int)cParams.targetLength); |
991 | 0 | BOUNDCHECK(ZSTD_c_strategy, cParams.strategy); |
992 | 0 | return 0; |
993 | 0 | } |
994 | | |
995 | | /** ZSTD_clampCParams() : |
996 | | * make CParam values within valid range. |
997 | | * @return : valid CParams */ |
998 | | static ZSTD_compressionParameters |
999 | | ZSTD_clampCParams(ZSTD_compressionParameters cParams) |
1000 | 0 | { |
1001 | 0 | # define CLAMP_TYPE(cParam, val, type) { \ |
1002 | 0 | ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam); \ |
1003 | 0 | if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound; \ |
1004 | 0 | else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \ |
1005 | 0 | } |
1006 | 0 | # define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, unsigned) |
1007 | 0 | CLAMP(ZSTD_c_windowLog, cParams.windowLog); |
1008 | 0 | CLAMP(ZSTD_c_chainLog, cParams.chainLog); |
1009 | 0 | CLAMP(ZSTD_c_hashLog, cParams.hashLog); |
1010 | 0 | CLAMP(ZSTD_c_searchLog, cParams.searchLog); |
1011 | 0 | CLAMP(ZSTD_c_minMatch, cParams.minMatch); |
1012 | 0 | CLAMP(ZSTD_c_targetLength,cParams.targetLength); |
1013 | 0 | CLAMP_TYPE(ZSTD_c_strategy,cParams.strategy, ZSTD_strategy); |
1014 | 0 | return cParams; |
1015 | 0 | } |
1016 | | |
1017 | | /** ZSTD_cycleLog() : |
1018 | | * condition for correct operation : hashLog > 1 */ |
1019 | | U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat) |
1020 | 0 | { |
1021 | 0 | U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2); |
1022 | 0 | return hashLog - btScale; |
1023 | 0 | } |
1024 | | |
1025 | | /** ZSTD_adjustCParams_internal() : |
1026 | | * optimize `cPar` for a specified input (`srcSize` and `dictSize`). |
1027 | | * mostly downsize to reduce memory consumption and initialization latency. |
1028 | | * `srcSize` can be ZSTD_CONTENTSIZE_UNKNOWN when not known. |
1029 | | * note : `srcSize==0` means 0! |
1030 | | * condition : cPar is presumed validated (can be checked using ZSTD_checkCParams()). */ |
1031 | | static ZSTD_compressionParameters |
1032 | | ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar, |
1033 | | unsigned long long srcSize, |
1034 | | size_t dictSize) |
1035 | 0 | { |
1036 | 0 | static const U64 minSrcSize = 513; /* (1<<9) + 1 */ |
1037 | 0 | static const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1); |
1038 | 0 | assert(ZSTD_checkCParams(cPar)==0); |
1039 | |
|
1040 | 0 | if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN) |
1041 | 0 | srcSize = minSrcSize; |
1042 | | |
1043 | | /* resize windowLog if input is small enough, to use less memory */ |
1044 | 0 | if ( (srcSize < maxWindowResize) |
1045 | 0 | && (dictSize < maxWindowResize) ) { |
1046 | 0 | U32 const tSize = (U32)(srcSize + dictSize); |
1047 | 0 | static U32 const hashSizeMin = 1 << ZSTD_HASHLOG_MIN; |
1048 | 0 | U32 const srcLog = (tSize < hashSizeMin) ? ZSTD_HASHLOG_MIN : |
1049 | 0 | ZSTD_highbit32(tSize-1) + 1; |
1050 | 0 | if (cPar.windowLog > srcLog) cPar.windowLog = srcLog; |
1051 | 0 | } |
1052 | 0 | if (cPar.hashLog > cPar.windowLog+1) cPar.hashLog = cPar.windowLog+1; |
1053 | 0 | { U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy); |
1054 | 0 | if (cycleLog > cPar.windowLog) |
1055 | 0 | cPar.chainLog -= (cycleLog - cPar.windowLog); |
1056 | 0 | } |
1057 | |
|
1058 | 0 | if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN) |
1059 | 0 | cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN; /* minimum wlog required for valid frame header */ |
1060 | |
|
1061 | 0 | return cPar; |
1062 | 0 | } |
1063 | | |
1064 | | ZSTD_compressionParameters |
1065 | | ZSTD_adjustCParams(ZSTD_compressionParameters cPar, |
1066 | | unsigned long long srcSize, |
1067 | | size_t dictSize) |
1068 | 0 | { |
1069 | 0 | cPar = ZSTD_clampCParams(cPar); /* resulting cPar is necessarily valid (all parameters within range) */ |
1070 | 0 | if (srcSize == 0) srcSize = ZSTD_CONTENTSIZE_UNKNOWN; |
1071 | 0 | return ZSTD_adjustCParams_internal(cPar, srcSize, dictSize); |
1072 | 0 | } |
1073 | | |
1074 | | static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize); |
1075 | | static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize); |
1076 | | |
1077 | | ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams( |
1078 | | const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize) |
1079 | 0 | { |
1080 | 0 | ZSTD_compressionParameters cParams; |
1081 | 0 | if (srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN && CCtxParams->srcSizeHint > 0) { |
1082 | 0 | srcSizeHint = CCtxParams->srcSizeHint; |
1083 | 0 | } |
1084 | 0 | cParams = ZSTD_getCParams_internal(CCtxParams->compressionLevel, srcSizeHint, dictSize); |
1085 | 0 | if (CCtxParams->ldmParams.enableLdm) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG; |
1086 | 0 | if (CCtxParams->cParams.windowLog) cParams.windowLog = CCtxParams->cParams.windowLog; |
1087 | 0 | if (CCtxParams->cParams.hashLog) cParams.hashLog = CCtxParams->cParams.hashLog; |
1088 | 0 | if (CCtxParams->cParams.chainLog) cParams.chainLog = CCtxParams->cParams.chainLog; |
1089 | 0 | if (CCtxParams->cParams.searchLog) cParams.searchLog = CCtxParams->cParams.searchLog; |
1090 | 0 | if (CCtxParams->cParams.minMatch) cParams.minMatch = CCtxParams->cParams.minMatch; |
1091 | 0 | if (CCtxParams->cParams.targetLength) cParams.targetLength = CCtxParams->cParams.targetLength; |
1092 | 0 | if (CCtxParams->cParams.strategy) cParams.strategy = CCtxParams->cParams.strategy; |
1093 | 0 | assert(!ZSTD_checkCParams(cParams)); |
1094 | | /* srcSizeHint == 0 means 0 */ |
1095 | 0 | return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize); |
1096 | 0 | } |
1097 | | |
1098 | | static size_t |
1099 | | ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams, |
1100 | | const U32 forCCtx) |
1101 | 0 | { |
1102 | 0 | size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog); |
1103 | 0 | size_t const hSize = ((size_t)1) << cParams->hashLog; |
1104 | 0 | U32 const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0; |
1105 | 0 | size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0; |
1106 | | /* We don't use ZSTD_cwksp_alloc_size() here because the tables aren't |
1107 | | * surrounded by redzones in ASAN. */ |
1108 | 0 | size_t const tableSpace = chainSize * sizeof(U32) |
1109 | 0 | + hSize * sizeof(U32) |
1110 | 0 | + h3Size * sizeof(U32); |
1111 | 0 | size_t const optPotentialSpace = |
1112 | 0 | ZSTD_cwksp_alloc_size((MaxML+1) * sizeof(U32)) |
1113 | 0 | + ZSTD_cwksp_alloc_size((MaxLL+1) * sizeof(U32)) |
1114 | 0 | + ZSTD_cwksp_alloc_size((MaxOff+1) * sizeof(U32)) |
1115 | 0 | + ZSTD_cwksp_alloc_size((1<<Litbits) * sizeof(U32)) |
1116 | 0 | + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t)) |
1117 | 0 | + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t)); |
1118 | 0 | size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt)) |
1119 | 0 | ? optPotentialSpace |
1120 | 0 | : 0; |
1121 | 0 | DEBUGLOG(4, "chainSize: %u - hSize: %u - h3Size: %u", |
1122 | 0 | (U32)chainSize, (U32)hSize, (U32)h3Size); |
1123 | 0 | return tableSpace + optSpace; |
1124 | 0 | } |
1125 | | |
1126 | | size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params) |
1127 | 0 | { |
1128 | 0 | RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only."); |
1129 | 0 | { ZSTD_compressionParameters const cParams = |
1130 | 0 | ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0); |
1131 | 0 | size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog); |
1132 | 0 | U32 const divider = (cParams.minMatch==3) ? 3 : 4; |
1133 | 0 | size_t const maxNbSeq = blockSize / divider; |
1134 | 0 | size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize) |
1135 | 0 | + ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef)) |
1136 | 0 | + 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE)); |
1137 | 0 | size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE); |
1138 | 0 | size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t)); |
1139 | 0 | size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 1); |
1140 | |
|
1141 | 0 | size_t const ldmSpace = ZSTD_ldm_getTableSize(params->ldmParams); |
1142 | 0 | size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize) * sizeof(rawSeq)); |
1143 | | |
1144 | | /* estimateCCtxSize is for one-shot compression. So no buffers should |
1145 | | * be needed. However, we still allocate two 0-sized buffers, which can |
1146 | | * take space under ASAN. */ |
1147 | 0 | size_t const bufferSpace = ZSTD_cwksp_alloc_size(0) |
1148 | 0 | + ZSTD_cwksp_alloc_size(0); |
1149 | |
|
1150 | 0 | size_t const cctxSpace = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)); |
1151 | |
|
1152 | 0 | size_t const neededSpace = |
1153 | 0 | cctxSpace + |
1154 | 0 | entropySpace + |
1155 | 0 | blockStateSpace + |
1156 | 0 | ldmSpace + |
1157 | 0 | ldmSeqSpace + |
1158 | 0 | matchStateSize + |
1159 | 0 | tokenSpace + |
1160 | 0 | bufferSpace; |
1161 | |
|
1162 | 0 | DEBUGLOG(5, "estimate workspace : %u", (U32)neededSpace); |
1163 | 0 | return neededSpace; |
1164 | 0 | } |
1165 | 0 | } |
1166 | | |
1167 | | size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams) |
1168 | 0 | { |
1169 | 0 | ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams); |
1170 | 0 | return ZSTD_estimateCCtxSize_usingCCtxParams(¶ms); |
1171 | 0 | } |
1172 | | |
1173 | | static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel) |
1174 | 0 | { |
1175 | 0 | ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0); |
1176 | 0 | return ZSTD_estimateCCtxSize_usingCParams(cParams); |
1177 | 0 | } |
1178 | | |
1179 | | size_t ZSTD_estimateCCtxSize(int compressionLevel) |
1180 | 0 | { |
1181 | 0 | int level; |
1182 | 0 | size_t memBudget = 0; |
1183 | 0 | for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) { |
1184 | 0 | size_t const newMB = ZSTD_estimateCCtxSize_internal(level); |
1185 | 0 | if (newMB > memBudget) memBudget = newMB; |
1186 | 0 | } |
1187 | 0 | return memBudget; |
1188 | 0 | } |
1189 | | |
1190 | | size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params) |
1191 | 0 | { |
1192 | 0 | RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only."); |
1193 | 0 | { ZSTD_compressionParameters const cParams = |
1194 | 0 | ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0); |
1195 | 0 | size_t const CCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(params); |
1196 | 0 | size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog); |
1197 | 0 | size_t const inBuffSize = ((size_t)1 << cParams.windowLog) + blockSize; |
1198 | 0 | size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1; |
1199 | 0 | size_t const streamingSize = ZSTD_cwksp_alloc_size(inBuffSize) |
1200 | 0 | + ZSTD_cwksp_alloc_size(outBuffSize); |
1201 | |
|
1202 | 0 | return CCtxSize + streamingSize; |
1203 | 0 | } |
1204 | 0 | } |
1205 | | |
1206 | | size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams) |
1207 | 0 | { |
1208 | 0 | ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams); |
1209 | 0 | return ZSTD_estimateCStreamSize_usingCCtxParams(¶ms); |
1210 | 0 | } |
1211 | | |
1212 | | static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel) |
1213 | 0 | { |
1214 | 0 | ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0); |
1215 | 0 | return ZSTD_estimateCStreamSize_usingCParams(cParams); |
1216 | 0 | } |
1217 | | |
1218 | | size_t ZSTD_estimateCStreamSize(int compressionLevel) |
1219 | 0 | { |
1220 | 0 | int level; |
1221 | 0 | size_t memBudget = 0; |
1222 | 0 | for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) { |
1223 | 0 | size_t const newMB = ZSTD_estimateCStreamSize_internal(level); |
1224 | 0 | if (newMB > memBudget) memBudget = newMB; |
1225 | 0 | } |
1226 | 0 | return memBudget; |
1227 | 0 | } |
1228 | | |
1229 | | /* ZSTD_getFrameProgression(): |
1230 | | * tells how much data has been consumed (input) and produced (output) for current frame. |
1231 | | * able to count progression inside worker threads (non-blocking mode). |
1232 | | */ |
1233 | | ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx) |
1234 | 0 | { |
1235 | | #ifdef ZSTD_MULTITHREAD |
1236 | | if (cctx->appliedParams.nbWorkers > 0) { |
1237 | | return ZSTDMT_getFrameProgression(cctx->mtctx); |
1238 | | } |
1239 | | #endif |
1240 | 0 | { ZSTD_frameProgression fp; |
1241 | 0 | size_t const buffered = (cctx->inBuff == NULL) ? 0 : |
1242 | 0 | cctx->inBuffPos - cctx->inToCompress; |
1243 | 0 | if (buffered) assert(cctx->inBuffPos >= cctx->inToCompress); |
1244 | 0 | assert(buffered <= ZSTD_BLOCKSIZE_MAX); |
1245 | 0 | fp.ingested = cctx->consumedSrcSize + buffered; |
1246 | 0 | fp.consumed = cctx->consumedSrcSize; |
1247 | 0 | fp.produced = cctx->producedCSize; |
1248 | 0 | fp.flushed = cctx->producedCSize; /* simplified; some data might still be left within streaming output buffer */ |
1249 | 0 | fp.currentJobID = 0; |
1250 | 0 | fp.nbActiveWorkers = 0; |
1251 | 0 | return fp; |
1252 | 0 | } } |
1253 | | |
1254 | | /*! ZSTD_toFlushNow() |
1255 | | * Only useful for multithreading scenarios currently (nbWorkers >= 1). |
1256 | | */ |
1257 | | size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx) |
1258 | 0 | { |
1259 | | #ifdef ZSTD_MULTITHREAD |
1260 | | if (cctx->appliedParams.nbWorkers > 0) { |
1261 | | return ZSTDMT_toFlushNow(cctx->mtctx); |
1262 | | } |
1263 | | #endif |
1264 | 0 | (void)cctx; |
1265 | 0 | return 0; /* over-simplification; could also check if context is currently running in streaming mode, and in which case, report how many bytes are left to be flushed within output buffer */ |
1266 | 0 | } |
1267 | | |
1268 | | static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1, |
1269 | | ZSTD_compressionParameters cParams2) |
1270 | 0 | { |
1271 | 0 | (void)cParams1; |
1272 | 0 | (void)cParams2; |
1273 | 0 | assert(cParams1.windowLog == cParams2.windowLog); |
1274 | 0 | assert(cParams1.chainLog == cParams2.chainLog); |
1275 | 0 | assert(cParams1.hashLog == cParams2.hashLog); |
1276 | 0 | assert(cParams1.searchLog == cParams2.searchLog); |
1277 | 0 | assert(cParams1.minMatch == cParams2.minMatch); |
1278 | 0 | assert(cParams1.targetLength == cParams2.targetLength); |
1279 | 0 | assert(cParams1.strategy == cParams2.strategy); |
1280 | 0 | } |
1281 | | |
1282 | | void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs) |
1283 | 0 | { |
1284 | 0 | int i; |
1285 | 0 | for (i = 0; i < ZSTD_REP_NUM; ++i) |
1286 | 0 | bs->rep[i] = ZSTDInternalConstants::repStartValue[i]; |
1287 | 0 | bs->entropy.huf.repeatMode = HUF_repeat_none; |
1288 | 0 | bs->entropy.fse.offcode_repeatMode = FSE_repeat_none; |
1289 | 0 | bs->entropy.fse.matchlength_repeatMode = FSE_repeat_none; |
1290 | 0 | bs->entropy.fse.litlength_repeatMode = FSE_repeat_none; |
1291 | 0 | } |
1292 | | |
1293 | | /*! ZSTD_invalidateMatchState() |
1294 | | * Invalidate all the matches in the match finder tables. |
1295 | | * Requires nextSrc and base to be set (can be NULL). |
1296 | | */ |
1297 | | static void ZSTD_invalidateMatchState(ZSTD_matchState_t* ms) |
1298 | 0 | { |
1299 | 0 | ZSTD_window_clear(&ms->window); |
1300 | |
|
1301 | 0 | ms->nextToUpdate = ms->window.dictLimit; |
1302 | 0 | ms->loadedDictEnd = 0; |
1303 | 0 | ms->opt.litLengthSum = 0; /* force reset of btopt stats */ |
1304 | 0 | ms->dictMatchState = NULL; |
1305 | 0 | } |
1306 | | |
1307 | | /** |
1308 | | * Indicates whether this compression proceeds directly from user-provided |
1309 | | * source buffer to user-provided destination buffer (ZSTDb_not_buffered), or |
1310 | | * whether the context needs to buffer the input/output (ZSTDb_buffered). |
1311 | | */ |
1312 | | typedef enum { |
1313 | | ZSTDb_not_buffered, |
1314 | | ZSTDb_buffered |
1315 | | } ZSTD_buffered_policy_e; |
1316 | | |
1317 | | /** |
1318 | | * Controls, for this matchState reset, whether the tables need to be cleared / |
1319 | | * prepared for the coming compression (ZSTDcrp_makeClean), or whether the |
1320 | | * tables can be left unclean (ZSTDcrp_leaveDirty), because we know that a |
1321 | | * subsequent operation will overwrite the table space anyways (e.g., copying |
1322 | | * the matchState contents in from a CDict). |
1323 | | */ |
1324 | | typedef enum { |
1325 | | ZSTDcrp_makeClean, |
1326 | | ZSTDcrp_leaveDirty |
1327 | | } ZSTD_compResetPolicy_e; |
1328 | | |
1329 | | /** |
1330 | | * Controls, for this matchState reset, whether indexing can continue where it |
1331 | | * left off (ZSTDirp_continue), or whether it needs to be restarted from zero |
1332 | | * (ZSTDirp_reset). |
1333 | | */ |
1334 | | typedef enum { |
1335 | | ZSTDirp_continue, |
1336 | | ZSTDirp_reset |
1337 | | } ZSTD_indexResetPolicy_e; |
1338 | | |
1339 | | typedef enum { |
1340 | | ZSTD_resetTarget_CDict, |
1341 | | ZSTD_resetTarget_CCtx |
1342 | | } ZSTD_resetTarget_e; |
1343 | | |
1344 | | static size_t |
1345 | | ZSTD_reset_matchState(ZSTD_matchState_t* ms, |
1346 | | ZSTD_cwksp* ws, |
1347 | | const ZSTD_compressionParameters* cParams, |
1348 | | const ZSTD_compResetPolicy_e crp, |
1349 | | const ZSTD_indexResetPolicy_e forceResetIndex, |
1350 | | const ZSTD_resetTarget_e forWho) |
1351 | 0 | { |
1352 | 0 | size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog); |
1353 | 0 | size_t const hSize = ((size_t)1) << cParams->hashLog; |
1354 | 0 | U32 const hashLog3 = ((forWho == ZSTD_resetTarget_CCtx) && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0; |
1355 | 0 | size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0; |
1356 | |
|
1357 | 0 | DEBUGLOG(4, "reset indices : %u", forceResetIndex == ZSTDirp_reset); |
1358 | 0 | if (forceResetIndex == ZSTDirp_reset) { |
1359 | 0 | ZSTD_window_init(&ms->window); |
1360 | 0 | ZSTD_cwksp_mark_tables_dirty(ws); |
1361 | 0 | } |
1362 | |
|
1363 | 0 | ms->hashLog3 = hashLog3; |
1364 | |
|
1365 | 0 | ZSTD_invalidateMatchState(ms); |
1366 | |
|
1367 | 0 | assert(!ZSTD_cwksp_reserve_failed(ws)); /* check that allocation hasn't already failed */ |
1368 | |
|
1369 | 0 | ZSTD_cwksp_clear_tables(ws); |
1370 | |
|
1371 | 0 | DEBUGLOG(5, "reserving table space"); |
1372 | | /* table Space */ |
1373 | 0 | ms->hashTable = (U32*)ZSTD_cwksp_reserve_table(ws, hSize * sizeof(U32)); |
1374 | 0 | ms->chainTable = (U32*)ZSTD_cwksp_reserve_table(ws, chainSize * sizeof(U32)); |
1375 | 0 | ms->hashTable3 = (U32*)ZSTD_cwksp_reserve_table(ws, h3Size * sizeof(U32)); |
1376 | 0 | RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation, |
1377 | 0 | "failed a workspace allocation in ZSTD_reset_matchState"); |
1378 | |
|
1379 | 0 | DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_leaveDirty); |
1380 | 0 | if (crp!=ZSTDcrp_leaveDirty) { |
1381 | | /* reset tables only */ |
1382 | 0 | ZSTD_cwksp_clean_tables(ws); |
1383 | 0 | } |
1384 | | |
1385 | | /* opt parser space */ |
1386 | 0 | if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) { |
1387 | 0 | DEBUGLOG(4, "reserving optimal parser space"); |
1388 | 0 | ms->opt.litFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (1<<Litbits) * sizeof(unsigned)); |
1389 | 0 | ms->opt.litLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxLL+1) * sizeof(unsigned)); |
1390 | 0 | ms->opt.matchLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxML+1) * sizeof(unsigned)); |
1391 | 0 | ms->opt.offCodeFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxOff+1) * sizeof(unsigned)); |
1392 | 0 | ms->opt.matchTable = (ZSTD_match_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t)); |
1393 | 0 | ms->opt.priceTable = (ZSTD_optimal_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t)); |
1394 | 0 | } |
1395 | |
|
1396 | 0 | ms->cParams = *cParams; |
1397 | |
|
1398 | 0 | RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation, |
1399 | 0 | "failed a workspace allocation in ZSTD_reset_matchState"); |
1400 | |
|
1401 | 0 | return 0; |
1402 | 0 | } |
1403 | | |
1404 | | /* ZSTD_indexTooCloseToMax() : |
1405 | | * minor optimization : prefer memset() rather than reduceIndex() |
1406 | | * which is measurably slow in some circumstances (reported for Visual Studio). |
1407 | | * Works when re-using a context for a lot of smallish inputs : |
1408 | | * if all inputs are smaller than ZSTD_INDEXOVERFLOW_MARGIN, |
1409 | | * memset() will be triggered before reduceIndex(). |
1410 | | */ |
1411 | 0 | #define ZSTD_INDEXOVERFLOW_MARGIN (16 MB) |
1412 | | static int ZSTD_indexTooCloseToMax(ZSTD_window_t w) |
1413 | 0 | { |
1414 | 0 | return (size_t)(w.nextSrc - w.base) > (ZSTD_CURRENT_MAX - ZSTD_INDEXOVERFLOW_MARGIN); |
1415 | 0 | } |
1416 | | |
1417 | | /*! ZSTD_resetCCtx_internal() : |
1418 | | note : `params` are assumed fully validated at this stage */ |
1419 | | static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc, |
1420 | | ZSTD_CCtx_params params, |
1421 | | U64 const pledgedSrcSize, |
1422 | | ZSTD_compResetPolicy_e const crp, |
1423 | | ZSTD_buffered_policy_e const zbuff) |
1424 | 0 | { |
1425 | 0 | ZSTD_cwksp* const ws = &zc->workspace; |
1426 | 0 | DEBUGLOG(4, "ZSTD_resetCCtx_internal: pledgedSrcSize=%u, wlog=%u", |
1427 | 0 | (U32)pledgedSrcSize, params.cParams.windowLog); |
1428 | 0 | assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams))); |
1429 | |
|
1430 | 0 | zc->isFirstBlock = 1; |
1431 | |
|
1432 | 0 | if (params.ldmParams.enableLdm) { |
1433 | | /* Adjust long distance matching parameters */ |
1434 | 0 | ZSTD_ldm_adjustParameters(¶ms.ldmParams, ¶ms.cParams); |
1435 | 0 | assert(params.ldmParams.hashLog >= params.ldmParams.bucketSizeLog); |
1436 | 0 | assert(params.ldmParams.hashRateLog < 32); |
1437 | 0 | zc->ldmState.hashPower = ZSTD_rollingHash_primePower(params.ldmParams.minMatchLength); |
1438 | 0 | } |
1439 | |
|
1440 | 0 | { size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize)); |
1441 | 0 | size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize); |
1442 | 0 | U32 const divider = (params.cParams.minMatch==3) ? 3 : 4; |
1443 | 0 | size_t const maxNbSeq = blockSize / divider; |
1444 | 0 | size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize) |
1445 | 0 | + ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef)) |
1446 | 0 | + 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE)); |
1447 | 0 | size_t const buffOutSize = (zbuff==ZSTDb_buffered) ? ZSTD_compressBound(blockSize)+1 : 0; |
1448 | 0 | size_t const buffInSize = (zbuff==ZSTDb_buffered) ? windowSize + blockSize : 0; |
1449 | 0 | size_t const matchStateSize = ZSTD_sizeof_matchState(¶ms.cParams, /* forCCtx */ 1); |
1450 | 0 | size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params.ldmParams, blockSize); |
1451 | |
|
1452 | 0 | ZSTD_indexResetPolicy_e needsIndexReset = zc->initialized ? ZSTDirp_continue : ZSTDirp_reset; |
1453 | |
|
1454 | 0 | if (ZSTD_indexTooCloseToMax(zc->blockState.matchState.window)) { |
1455 | 0 | needsIndexReset = ZSTDirp_reset; |
1456 | 0 | } |
1457 | |
|
1458 | 0 | if (!zc->staticSize) ZSTD_cwksp_bump_oversized_duration(ws, 0); |
1459 | | |
1460 | | /* Check if workspace is large enough, alloc a new one if needed */ |
1461 | 0 | { size_t const cctxSpace = zc->staticSize ? ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)) : 0; |
1462 | 0 | size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE); |
1463 | 0 | size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t)); |
1464 | 0 | size_t const bufferSpace = ZSTD_cwksp_alloc_size(buffInSize) + ZSTD_cwksp_alloc_size(buffOutSize); |
1465 | 0 | size_t const ldmSpace = ZSTD_ldm_getTableSize(params.ldmParams); |
1466 | 0 | size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(maxNbLdmSeq * sizeof(rawSeq)); |
1467 | |
|
1468 | 0 | size_t const neededSpace = |
1469 | 0 | cctxSpace + |
1470 | 0 | entropySpace + |
1471 | 0 | blockStateSpace + |
1472 | 0 | ldmSpace + |
1473 | 0 | ldmSeqSpace + |
1474 | 0 | matchStateSize + |
1475 | 0 | tokenSpace + |
1476 | 0 | bufferSpace; |
1477 | |
|
1478 | 0 | int const workspaceTooSmall = ZSTD_cwksp_sizeof(ws) < neededSpace; |
1479 | 0 | int const workspaceWasteful = ZSTD_cwksp_check_wasteful(ws, neededSpace); |
1480 | |
|
1481 | 0 | DEBUGLOG(4, "Need %zuKB workspace, including %zuKB for match state, and %zuKB for buffers", |
1482 | 0 | neededSpace>>10, matchStateSize>>10, bufferSpace>>10); |
1483 | 0 | DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize); |
1484 | |
|
1485 | 0 | if (workspaceTooSmall || workspaceWasteful) { |
1486 | 0 | DEBUGLOG(4, "Resize workspaceSize from %zuKB to %zuKB", |
1487 | 0 | ZSTD_cwksp_sizeof(ws) >> 10, |
1488 | 0 | neededSpace >> 10); |
1489 | |
|
1490 | 0 | RETURN_ERROR_IF(zc->staticSize, memory_allocation, "static cctx : no resize"); |
1491 | |
|
1492 | 0 | needsIndexReset = ZSTDirp_reset; |
1493 | |
|
1494 | 0 | ZSTD_cwksp_free(ws, zc->customMem); |
1495 | 0 | FORWARD_IF_ERROR(ZSTD_cwksp_create(ws, neededSpace, zc->customMem), ""); |
1496 | |
|
1497 | 0 | DEBUGLOG(5, "reserving object space"); |
1498 | | /* Statically sized space. |
1499 | | * entropyWorkspace never moves, |
1500 | | * though prev/next block swap places */ |
1501 | 0 | assert(ZSTD_cwksp_check_available(ws, 2 * sizeof(ZSTD_compressedBlockState_t))); |
1502 | 0 | zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t)); |
1503 | 0 | RETURN_ERROR_IF(zc->blockState.prevCBlock == NULL, memory_allocation, "couldn't allocate prevCBlock"); |
1504 | 0 | zc->blockState.nextCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t)); |
1505 | 0 | RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate nextCBlock"); |
1506 | 0 | zc->entropyWorkspace = (U32*) ZSTD_cwksp_reserve_object(ws, HUF_WORKSPACE_SIZE); |
1507 | 0 | RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate entropyWorkspace"); |
1508 | 0 | } } |
1509 | | |
1510 | 0 | ZSTD_cwksp_clear(ws); |
1511 | | |
1512 | | /* init params */ |
1513 | 0 | zc->appliedParams = params; |
1514 | 0 | zc->blockState.matchState.cParams = params.cParams; |
1515 | 0 | zc->pledgedSrcSizePlusOne = pledgedSrcSize+1; |
1516 | 0 | zc->consumedSrcSize = 0; |
1517 | 0 | zc->producedCSize = 0; |
1518 | 0 | if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN) |
1519 | 0 | zc->appliedParams.fParams.contentSizeFlag = 0; |
1520 | 0 | DEBUGLOG(4, "pledged content size : %u ; flag : %u", |
1521 | 0 | (unsigned)pledgedSrcSize, zc->appliedParams.fParams.contentSizeFlag); |
1522 | 0 | zc->blockSize = blockSize; |
1523 | |
|
1524 | 0 | XXH64_reset(&zc->xxhState, 0); |
1525 | 0 | zc->stage = ZSTDcs_init; |
1526 | 0 | zc->dictID = 0; |
1527 | |
|
1528 | 0 | ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock); |
1529 | | |
1530 | | /* ZSTD_wildcopy() is used to copy into the literals buffer, |
1531 | | * so we have to oversize the buffer by WILDCOPY_OVERLENGTH bytes. |
1532 | | */ |
1533 | 0 | zc->seqStore.litStart = ZSTD_cwksp_reserve_buffer(ws, blockSize + WILDCOPY_OVERLENGTH); |
1534 | 0 | zc->seqStore.maxNbLit = blockSize; |
1535 | | |
1536 | | /* buffers */ |
1537 | 0 | zc->inBuffSize = buffInSize; |
1538 | 0 | zc->inBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffInSize); |
1539 | 0 | zc->outBuffSize = buffOutSize; |
1540 | 0 | zc->outBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffOutSize); |
1541 | | |
1542 | | /* ldm bucketOffsets table */ |
1543 | 0 | if (params.ldmParams.enableLdm) { |
1544 | | /* TODO: avoid memset? */ |
1545 | 0 | size_t const ldmBucketSize = |
1546 | 0 | ((size_t)1) << (params.ldmParams.hashLog - |
1547 | 0 | params.ldmParams.bucketSizeLog); |
1548 | 0 | zc->ldmState.bucketOffsets = ZSTD_cwksp_reserve_buffer(ws, ldmBucketSize); |
1549 | 0 | memset(zc->ldmState.bucketOffsets, 0, ldmBucketSize); |
1550 | 0 | } |
1551 | | |
1552 | | /* sequences storage */ |
1553 | 0 | ZSTD_referenceExternalSequences(zc, NULL, 0); |
1554 | 0 | zc->seqStore.maxNbSeq = maxNbSeq; |
1555 | 0 | zc->seqStore.llCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE)); |
1556 | 0 | zc->seqStore.mlCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE)); |
1557 | 0 | zc->seqStore.ofCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE)); |
1558 | 0 | zc->seqStore.sequencesStart = (seqDef*)ZSTD_cwksp_reserve_aligned(ws, maxNbSeq * sizeof(seqDef)); |
1559 | |
|
1560 | 0 | FORWARD_IF_ERROR(ZSTD_reset_matchState( |
1561 | 0 | &zc->blockState.matchState, |
1562 | 0 | ws, |
1563 | 0 | ¶ms.cParams, |
1564 | 0 | crp, |
1565 | 0 | needsIndexReset, |
1566 | 0 | ZSTD_resetTarget_CCtx), ""); |
1567 | | |
1568 | | /* ldm hash table */ |
1569 | 0 | if (params.ldmParams.enableLdm) { |
1570 | | /* TODO: avoid memset? */ |
1571 | 0 | size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog; |
1572 | 0 | zc->ldmState.hashTable = (ldmEntry_t*)ZSTD_cwksp_reserve_aligned(ws, ldmHSize * sizeof(ldmEntry_t)); |
1573 | 0 | memset(zc->ldmState.hashTable, 0, ldmHSize * sizeof(ldmEntry_t)); |
1574 | 0 | zc->ldmSequences = (rawSeq*)ZSTD_cwksp_reserve_aligned(ws, maxNbLdmSeq * sizeof(rawSeq)); |
1575 | 0 | zc->maxNbLdmSequences = maxNbLdmSeq; |
1576 | |
|
1577 | 0 | ZSTD_window_init(&zc->ldmState.window); |
1578 | 0 | ZSTD_window_clear(&zc->ldmState.window); |
1579 | 0 | zc->ldmState.loadedDictEnd = 0; |
1580 | 0 | } |
1581 | |
|
1582 | 0 | DEBUGLOG(3, "wksp: finished allocating, %zd bytes remain available", ZSTD_cwksp_available_space(ws)); |
1583 | 0 | zc->initialized = 1; |
1584 | |
|
1585 | 0 | return 0; |
1586 | 0 | } |
1587 | 0 | } |
1588 | | |
1589 | | /* ZSTD_invalidateRepCodes() : |
1590 | | * ensures next compression will not use repcodes from previous block. |
1591 | | * Note : only works with regular variant; |
1592 | | * do not use with extDict variant ! */ |
1593 | 0 | void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx) { |
1594 | 0 | int i; |
1595 | 0 | for (i=0; i<ZSTD_REP_NUM; i++) cctx->blockState.prevCBlock->rep[i] = 0; |
1596 | 0 | assert(!ZSTD_window_hasExtDict(cctx->blockState.matchState.window)); |
1597 | 0 | } |
1598 | | |
1599 | | /* These are the approximate sizes for each strategy past which copying the |
1600 | | * dictionary tables into the working context is faster than using them |
1601 | | * in-place. |
1602 | | */ |
1603 | | static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = { |
1604 | | 8 KB, /* unused */ |
1605 | | 8 KB, /* ZSTD_fast */ |
1606 | | 16 KB, /* ZSTD_dfast */ |
1607 | | 32 KB, /* ZSTD_greedy */ |
1608 | | 32 KB, /* ZSTD_lazy */ |
1609 | | 32 KB, /* ZSTD_lazy2 */ |
1610 | | 32 KB, /* ZSTD_btlazy2 */ |
1611 | | 32 KB, /* ZSTD_btopt */ |
1612 | | 8 KB, /* ZSTD_btultra */ |
1613 | | 8 KB /* ZSTD_btultra2 */ |
1614 | | }; |
1615 | | |
1616 | | static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict, |
1617 | | const ZSTD_CCtx_params* params, |
1618 | | U64 pledgedSrcSize) |
1619 | 0 | { |
1620 | 0 | size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy]; |
1621 | 0 | return ( pledgedSrcSize <= cutoff |
1622 | 0 | || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN |
1623 | 0 | || params->attachDictPref == ZSTD_dictForceAttach ) |
1624 | 0 | && params->attachDictPref != ZSTD_dictForceCopy |
1625 | 0 | && !params->forceWindow; /* dictMatchState isn't correctly |
1626 | | * handled in _enforceMaxDist */ |
1627 | 0 | } |
1628 | | |
1629 | | static size_t |
1630 | | ZSTD_resetCCtx_byAttachingCDict(ZSTD_CCtx* cctx, |
1631 | | const ZSTD_CDict* cdict, |
1632 | | ZSTD_CCtx_params params, |
1633 | | U64 pledgedSrcSize, |
1634 | | ZSTD_buffered_policy_e zbuff) |
1635 | 0 | { |
1636 | 0 | { const ZSTD_compressionParameters* const cdict_cParams = &cdict->matchState.cParams; |
1637 | 0 | unsigned const windowLog = params.cParams.windowLog; |
1638 | 0 | assert(windowLog != 0); |
1639 | | /* Resize working context table params for input only, since the dict |
1640 | | * has its own tables. */ |
1641 | | /* pledgeSrcSize == 0 means 0! */ |
1642 | 0 | params.cParams = ZSTD_adjustCParams_internal(*cdict_cParams, pledgedSrcSize, 0); |
1643 | 0 | params.cParams.windowLog = windowLog; |
1644 | 0 | FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize, |
1645 | 0 | ZSTDcrp_makeClean, zbuff), ""); |
1646 | 0 | assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy); |
1647 | 0 | } |
1648 | | |
1649 | 0 | { const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc |
1650 | 0 | - cdict->matchState.window.base); |
1651 | 0 | const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit; |
1652 | 0 | if (cdictLen == 0) { |
1653 | | /* don't even attach dictionaries with no contents */ |
1654 | 0 | DEBUGLOG(4, "skipping attaching empty dictionary"); |
1655 | 0 | } else { |
1656 | 0 | DEBUGLOG(4, "attaching dictionary into context"); |
1657 | 0 | cctx->blockState.matchState.dictMatchState = &cdict->matchState; |
1658 | | |
1659 | | /* prep working match state so dict matches never have negative indices |
1660 | | * when they are translated to the working context's index space. */ |
1661 | 0 | if (cctx->blockState.matchState.window.dictLimit < cdictEnd) { |
1662 | 0 | cctx->blockState.matchState.window.nextSrc = |
1663 | 0 | cctx->blockState.matchState.window.base + cdictEnd; |
1664 | 0 | ZSTD_window_clear(&cctx->blockState.matchState.window); |
1665 | 0 | } |
1666 | | /* loadedDictEnd is expressed within the referential of the active context */ |
1667 | 0 | cctx->blockState.matchState.loadedDictEnd = cctx->blockState.matchState.window.dictLimit; |
1668 | 0 | } } |
1669 | |
|
1670 | 0 | cctx->dictID = cdict->dictID; |
1671 | | |
1672 | | /* copy block state */ |
1673 | 0 | memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState)); |
1674 | |
|
1675 | 0 | return 0; |
1676 | 0 | } |
1677 | | |
1678 | | static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx, |
1679 | | const ZSTD_CDict* cdict, |
1680 | | ZSTD_CCtx_params params, |
1681 | | U64 pledgedSrcSize, |
1682 | | ZSTD_buffered_policy_e zbuff) |
1683 | 0 | { |
1684 | 0 | const ZSTD_compressionParameters *cdict_cParams = &cdict->matchState.cParams; |
1685 | |
|
1686 | 0 | DEBUGLOG(4, "copying dictionary into context"); |
1687 | |
|
1688 | 0 | { unsigned const windowLog = params.cParams.windowLog; |
1689 | 0 | assert(windowLog != 0); |
1690 | | /* Copy only compression parameters related to tables. */ |
1691 | 0 | params.cParams = *cdict_cParams; |
1692 | 0 | params.cParams.windowLog = windowLog; |
1693 | 0 | FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize, |
1694 | 0 | ZSTDcrp_leaveDirty, zbuff), ""); |
1695 | 0 | assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy); |
1696 | 0 | assert(cctx->appliedParams.cParams.hashLog == cdict_cParams->hashLog); |
1697 | 0 | assert(cctx->appliedParams.cParams.chainLog == cdict_cParams->chainLog); |
1698 | 0 | } |
1699 | | |
1700 | 0 | ZSTD_cwksp_mark_tables_dirty(&cctx->workspace); |
1701 | | |
1702 | | /* copy tables */ |
1703 | 0 | { size_t const chainSize = (cdict_cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cdict_cParams->chainLog); |
1704 | 0 | size_t const hSize = (size_t)1 << cdict_cParams->hashLog; |
1705 | |
|
1706 | 0 | memcpy(cctx->blockState.matchState.hashTable, |
1707 | 0 | cdict->matchState.hashTable, |
1708 | 0 | hSize * sizeof(U32)); |
1709 | 0 | memcpy(cctx->blockState.matchState.chainTable, |
1710 | 0 | cdict->matchState.chainTable, |
1711 | 0 | chainSize * sizeof(U32)); |
1712 | 0 | } |
1713 | | |
1714 | | /* Zero the hashTable3, since the cdict never fills it */ |
1715 | 0 | { int const h3log = cctx->blockState.matchState.hashLog3; |
1716 | 0 | size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0; |
1717 | 0 | assert(cdict->matchState.hashLog3 == 0); |
1718 | 0 | memset(cctx->blockState.matchState.hashTable3, 0, h3Size * sizeof(U32)); |
1719 | 0 | } |
1720 | |
|
1721 | 0 | ZSTD_cwksp_mark_tables_clean(&cctx->workspace); |
1722 | | |
1723 | | /* copy dictionary offsets */ |
1724 | 0 | { ZSTD_matchState_t const* srcMatchState = &cdict->matchState; |
1725 | 0 | ZSTD_matchState_t* dstMatchState = &cctx->blockState.matchState; |
1726 | 0 | dstMatchState->window = srcMatchState->window; |
1727 | 0 | dstMatchState->nextToUpdate = srcMatchState->nextToUpdate; |
1728 | 0 | dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd; |
1729 | 0 | } |
1730 | |
|
1731 | 0 | cctx->dictID = cdict->dictID; |
1732 | | |
1733 | | /* copy block state */ |
1734 | 0 | memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState)); |
1735 | |
|
1736 | 0 | return 0; |
1737 | 0 | } |
1738 | | |
1739 | | /* We have a choice between copying the dictionary context into the working |
1740 | | * context, or referencing the dictionary context from the working context |
1741 | | * in-place. We decide here which strategy to use. */ |
1742 | | static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx, |
1743 | | const ZSTD_CDict* cdict, |
1744 | | const ZSTD_CCtx_params* params, |
1745 | | U64 pledgedSrcSize, |
1746 | | ZSTD_buffered_policy_e zbuff) |
1747 | 0 | { |
1748 | |
|
1749 | 0 | DEBUGLOG(4, "ZSTD_resetCCtx_usingCDict (pledgedSrcSize=%u)", |
1750 | 0 | (unsigned)pledgedSrcSize); |
1751 | |
|
1752 | 0 | if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) { |
1753 | 0 | return ZSTD_resetCCtx_byAttachingCDict( |
1754 | 0 | cctx, cdict, *params, pledgedSrcSize, zbuff); |
1755 | 0 | } else { |
1756 | 0 | return ZSTD_resetCCtx_byCopyingCDict( |
1757 | 0 | cctx, cdict, *params, pledgedSrcSize, zbuff); |
1758 | 0 | } |
1759 | 0 | } |
1760 | | |
1761 | | /*! ZSTD_copyCCtx_internal() : |
1762 | | * Duplicate an existing context `srcCCtx` into another one `dstCCtx`. |
1763 | | * Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()). |
1764 | | * The "context", in this case, refers to the hash and chain tables, |
1765 | | * entropy tables, and dictionary references. |
1766 | | * `windowLog` value is enforced if != 0, otherwise value is copied from srcCCtx. |
1767 | | * @return : 0, or an error code */ |
1768 | | static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx, |
1769 | | const ZSTD_CCtx* srcCCtx, |
1770 | | ZSTD_frameParameters fParams, |
1771 | | U64 pledgedSrcSize, |
1772 | | ZSTD_buffered_policy_e zbuff) |
1773 | 0 | { |
1774 | 0 | DEBUGLOG(5, "ZSTD_copyCCtx_internal"); |
1775 | 0 | RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong, |
1776 | 0 | "Can't copy a ctx that's not in init stage."); |
1777 | |
|
1778 | 0 | memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem)); |
1779 | 0 | { ZSTD_CCtx_params params = dstCCtx->requestedParams; |
1780 | | /* Copy only compression parameters related to tables. */ |
1781 | 0 | params.cParams = srcCCtx->appliedParams.cParams; |
1782 | 0 | params.fParams = fParams; |
1783 | 0 | ZSTD_resetCCtx_internal(dstCCtx, params, pledgedSrcSize, |
1784 | 0 | ZSTDcrp_leaveDirty, zbuff); |
1785 | 0 | assert(dstCCtx->appliedParams.cParams.windowLog == srcCCtx->appliedParams.cParams.windowLog); |
1786 | 0 | assert(dstCCtx->appliedParams.cParams.strategy == srcCCtx->appliedParams.cParams.strategy); |
1787 | 0 | assert(dstCCtx->appliedParams.cParams.hashLog == srcCCtx->appliedParams.cParams.hashLog); |
1788 | 0 | assert(dstCCtx->appliedParams.cParams.chainLog == srcCCtx->appliedParams.cParams.chainLog); |
1789 | 0 | assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3); |
1790 | 0 | } |
1791 | |
|
1792 | 0 | ZSTD_cwksp_mark_tables_dirty(&dstCCtx->workspace); |
1793 | | |
1794 | | /* copy tables */ |
1795 | 0 | { size_t const chainSize = (srcCCtx->appliedParams.cParams.strategy == ZSTD_fast) ? 0 : ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog); |
1796 | 0 | size_t const hSize = (size_t)1 << srcCCtx->appliedParams.cParams.hashLog; |
1797 | 0 | int const h3log = srcCCtx->blockState.matchState.hashLog3; |
1798 | 0 | size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0; |
1799 | |
|
1800 | 0 | memcpy(dstCCtx->blockState.matchState.hashTable, |
1801 | 0 | srcCCtx->blockState.matchState.hashTable, |
1802 | 0 | hSize * sizeof(U32)); |
1803 | 0 | memcpy(dstCCtx->blockState.matchState.chainTable, |
1804 | 0 | srcCCtx->blockState.matchState.chainTable, |
1805 | 0 | chainSize * sizeof(U32)); |
1806 | 0 | memcpy(dstCCtx->blockState.matchState.hashTable3, |
1807 | 0 | srcCCtx->blockState.matchState.hashTable3, |
1808 | 0 | h3Size * sizeof(U32)); |
1809 | 0 | } |
1810 | |
|
1811 | 0 | ZSTD_cwksp_mark_tables_clean(&dstCCtx->workspace); |
1812 | | |
1813 | | /* copy dictionary offsets */ |
1814 | 0 | { |
1815 | 0 | const ZSTD_matchState_t* srcMatchState = &srcCCtx->blockState.matchState; |
1816 | 0 | ZSTD_matchState_t* dstMatchState = &dstCCtx->blockState.matchState; |
1817 | 0 | dstMatchState->window = srcMatchState->window; |
1818 | 0 | dstMatchState->nextToUpdate = srcMatchState->nextToUpdate; |
1819 | 0 | dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd; |
1820 | 0 | } |
1821 | 0 | dstCCtx->dictID = srcCCtx->dictID; |
1822 | | |
1823 | | /* copy block state */ |
1824 | 0 | memcpy(dstCCtx->blockState.prevCBlock, srcCCtx->blockState.prevCBlock, sizeof(*srcCCtx->blockState.prevCBlock)); |
1825 | |
|
1826 | 0 | return 0; |
1827 | 0 | } |
1828 | | |
1829 | | /*! ZSTD_copyCCtx() : |
1830 | | * Duplicate an existing context `srcCCtx` into another one `dstCCtx`. |
1831 | | * Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()). |
1832 | | * pledgedSrcSize==0 means "unknown". |
1833 | | * @return : 0, or an error code */ |
1834 | | size_t ZSTD_copyCCtx(ZSTD_CCtx* dstCCtx, const ZSTD_CCtx* srcCCtx, unsigned long long pledgedSrcSize) |
1835 | 0 | { |
1836 | 0 | ZSTD_frameParameters fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ }; |
1837 | 0 | ZSTD_buffered_policy_e const zbuff = (ZSTD_buffered_policy_e)(srcCCtx->inBuffSize>0); |
1838 | 0 | ZSTD_STATIC_ASSERT((U32)ZSTDb_buffered==1); |
1839 | 0 | if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN; |
1840 | 0 | fParams.contentSizeFlag = (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN); |
1841 | |
|
1842 | 0 | return ZSTD_copyCCtx_internal(dstCCtx, srcCCtx, |
1843 | 0 | fParams, pledgedSrcSize, |
1844 | 0 | zbuff); |
1845 | 0 | } |
1846 | | |
1847 | | |
1848 | 0 | #define ZSTD_ROWSIZE 16 |
1849 | | /*! ZSTD_reduceTable() : |
1850 | | * reduce table indexes by `reducerValue`, or squash to zero. |
1851 | | * PreserveMark preserves "unsorted mark" for btlazy2 strategy. |
1852 | | * It must be set to a clear 0/1 value, to remove branch during inlining. |
1853 | | * Presume table size is a multiple of ZSTD_ROWSIZE |
1854 | | * to help auto-vectorization */ |
1855 | | FORCE_INLINE_TEMPLATE void |
1856 | | ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerValue, int const preserveMark) |
1857 | 0 | { |
1858 | 0 | int const nbRows = (int)size / ZSTD_ROWSIZE; |
1859 | 0 | int cellNb = 0; |
1860 | 0 | int rowNb; |
1861 | 0 | assert((size & (ZSTD_ROWSIZE-1)) == 0); /* multiple of ZSTD_ROWSIZE */ |
1862 | 0 | assert(size < (1U<<31)); /* can be casted to int */ |
1863 | |
|
1864 | | #if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE) |
1865 | | /* To validate that the table re-use logic is sound, and that we don't |
1866 | | * access table space that we haven't cleaned, we re-"poison" the table |
1867 | | * space every time we mark it dirty. |
1868 | | * |
1869 | | * This function however is intended to operate on those dirty tables and |
1870 | | * re-clean them. So when this function is used correctly, we can unpoison |
1871 | | * the memory it operated on. This introduces a blind spot though, since |
1872 | | * if we now try to operate on __actually__ poisoned memory, we will not |
1873 | | * detect that. */ |
1874 | | __msan_unpoison(table, size * sizeof(U32)); |
1875 | | #endif |
1876 | |
|
1877 | 0 | for (rowNb=0 ; rowNb < nbRows ; rowNb++) { |
1878 | 0 | int column; |
1879 | 0 | for (column=0; column<ZSTD_ROWSIZE; column++) { |
1880 | 0 | if (preserveMark) { |
1881 | 0 | U32 const adder = (table[cellNb] == ZSTD_DUBT_UNSORTED_MARK) ? reducerValue : 0; |
1882 | 0 | table[cellNb] += adder; |
1883 | 0 | } |
1884 | 0 | if (table[cellNb] < reducerValue) table[cellNb] = 0; |
1885 | 0 | else table[cellNb] -= reducerValue; |
1886 | 0 | cellNb++; |
1887 | 0 | } } |
1888 | 0 | } |
1889 | | |
1890 | | static void ZSTD_reduceTable(U32* const table, U32 const size, U32 const reducerValue) |
1891 | 0 | { |
1892 | 0 | ZSTD_reduceTable_internal(table, size, reducerValue, 0); |
1893 | 0 | } |
1894 | | |
1895 | | static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const reducerValue) |
1896 | 0 | { |
1897 | 0 | ZSTD_reduceTable_internal(table, size, reducerValue, 1); |
1898 | 0 | } |
1899 | | |
1900 | | /*! ZSTD_reduceIndex() : |
1901 | | * rescale all indexes to avoid future overflow (indexes are U32) */ |
1902 | | static void ZSTD_reduceIndex (ZSTD_matchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue) |
1903 | 0 | { |
1904 | 0 | { U32 const hSize = (U32)1 << params->cParams.hashLog; |
1905 | 0 | ZSTD_reduceTable(ms->hashTable, hSize, reducerValue); |
1906 | 0 | } |
1907 | |
|
1908 | 0 | if (params->cParams.strategy != ZSTD_fast) { |
1909 | 0 | U32 const chainSize = (U32)1 << params->cParams.chainLog; |
1910 | 0 | if (params->cParams.strategy == ZSTD_btlazy2) |
1911 | 0 | ZSTD_reduceTable_btlazy2(ms->chainTable, chainSize, reducerValue); |
1912 | 0 | else |
1913 | 0 | ZSTD_reduceTable(ms->chainTable, chainSize, reducerValue); |
1914 | 0 | } |
1915 | |
|
1916 | 0 | if (ms->hashLog3) { |
1917 | 0 | U32 const h3Size = (U32)1 << ms->hashLog3; |
1918 | 0 | ZSTD_reduceTable(ms->hashTable3, h3Size, reducerValue); |
1919 | 0 | } |
1920 | 0 | } |
1921 | | |
1922 | | |
1923 | | /*-******************************************************* |
1924 | | * Block entropic compression |
1925 | | *********************************************************/ |
1926 | | |
1927 | | /* See doc/zstd_compression_format.md for detailed format description */ |
1928 | | |
1929 | | void ZSTD_seqToCodes(const seqStore_t* seqStorePtr) |
1930 | 0 | { |
1931 | 0 | const seqDef* const sequences = seqStorePtr->sequencesStart; |
1932 | 0 | BYTE* const llCodeTable = seqStorePtr->llCode; |
1933 | 0 | BYTE* const ofCodeTable = seqStorePtr->ofCode; |
1934 | 0 | BYTE* const mlCodeTable = seqStorePtr->mlCode; |
1935 | 0 | U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart); |
1936 | 0 | U32 u; |
1937 | 0 | assert(nbSeq <= seqStorePtr->maxNbSeq); |
1938 | 0 | for (u=0; u<nbSeq; u++) { |
1939 | 0 | U32 const llv = sequences[u].litLength; |
1940 | 0 | U32 const mlv = sequences[u].matchLength; |
1941 | 0 | llCodeTable[u] = (BYTE)ZSTD_LLcode(llv); |
1942 | 0 | ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset); |
1943 | 0 | mlCodeTable[u] = (BYTE)ZSTD_MLcode(mlv); |
1944 | 0 | } |
1945 | 0 | if (seqStorePtr->longLengthID==1) |
1946 | 0 | llCodeTable[seqStorePtr->longLengthPos] = MaxLL; |
1947 | 0 | if (seqStorePtr->longLengthID==2) |
1948 | 0 | mlCodeTable[seqStorePtr->longLengthPos] = MaxML; |
1949 | 0 | } |
1950 | | |
1951 | | /* ZSTD_useTargetCBlockSize(): |
1952 | | * Returns if target compressed block size param is being used. |
1953 | | * If used, compression will do best effort to make a compressed block size to be around targetCBlockSize. |
1954 | | * Returns 1 if true, 0 otherwise. */ |
1955 | | static int ZSTD_useTargetCBlockSize(const ZSTD_CCtx_params* cctxParams) |
1956 | 0 | { |
1957 | 0 | DEBUGLOG(5, "ZSTD_useTargetCBlockSize (targetCBlockSize=%zu)", cctxParams->targetCBlockSize); |
1958 | 0 | return (cctxParams->targetCBlockSize != 0); |
1959 | 0 | } |
1960 | | |
1961 | | /* ZSTD_compressSequences_internal(): |
1962 | | * actually compresses both literals and sequences */ |
1963 | | MEM_STATIC size_t |
1964 | | ZSTD_compressSequences_internal(seqStore_t* seqStorePtr, |
1965 | | const ZSTD_entropyCTables_t* prevEntropy, |
1966 | | ZSTD_entropyCTables_t* nextEntropy, |
1967 | | const ZSTD_CCtx_params* cctxParams, |
1968 | | void* dst, size_t dstCapacity, |
1969 | | void* entropyWorkspace, size_t entropyWkspSize, |
1970 | | const int bmi2) |
1971 | 0 | { |
1972 | 0 | const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN; |
1973 | 0 | ZSTD_strategy const strategy = cctxParams->cParams.strategy; |
1974 | 0 | unsigned count[MaxSeq+1]; |
1975 | 0 | FSE_CTable* CTable_LitLength = nextEntropy->fse.litlengthCTable; |
1976 | 0 | FSE_CTable* CTable_OffsetBits = nextEntropy->fse.offcodeCTable; |
1977 | 0 | FSE_CTable* CTable_MatchLength = nextEntropy->fse.matchlengthCTable; |
1978 | 0 | U32 LLtype, Offtype, MLtype; /* compressed, raw or rle */ |
1979 | 0 | const seqDef* const sequences = seqStorePtr->sequencesStart; |
1980 | 0 | const BYTE* const ofCodeTable = seqStorePtr->ofCode; |
1981 | 0 | const BYTE* const llCodeTable = seqStorePtr->llCode; |
1982 | 0 | const BYTE* const mlCodeTable = seqStorePtr->mlCode; |
1983 | 0 | BYTE* const ostart = (BYTE*)dst; |
1984 | 0 | BYTE* const oend = ostart + dstCapacity; |
1985 | 0 | BYTE* op = ostart; |
1986 | 0 | size_t const nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart); |
1987 | 0 | BYTE* seqHead; |
1988 | 0 | BYTE* lastNCount = NULL; |
1989 | |
|
1990 | 0 | DEBUGLOG(5, "ZSTD_compressSequences_internal (nbSeq=%zu)", nbSeq); |
1991 | 0 | ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog))); |
1992 | | |
1993 | | /* Compress literals */ |
1994 | 0 | { const BYTE* const literals = seqStorePtr->litStart; |
1995 | 0 | size_t const litSize = (size_t)(seqStorePtr->lit - literals); |
1996 | 0 | size_t const cSize = ZSTD_compressLiterals( |
1997 | 0 | &prevEntropy->huf, &nextEntropy->huf, |
1998 | 0 | cctxParams->cParams.strategy, |
1999 | 0 | ZSTD_disableLiteralsCompression(cctxParams), |
2000 | 0 | op, dstCapacity, |
2001 | 0 | literals, litSize, |
2002 | 0 | entropyWorkspace, entropyWkspSize, |
2003 | 0 | bmi2); |
2004 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed"); |
2005 | 0 | assert(cSize <= dstCapacity); |
2006 | 0 | op += cSize; |
2007 | 0 | } |
2008 | | |
2009 | | /* Sequences Header */ |
2010 | 0 | RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/, |
2011 | 0 | dstSize_tooSmall, "Can't fit seq hdr in output buf!"); |
2012 | 0 | if (nbSeq < 128) { |
2013 | 0 | *op++ = (BYTE)nbSeq; |
2014 | 0 | } else if (nbSeq < LONGNBSEQ) { |
2015 | 0 | op[0] = (BYTE)((nbSeq>>8) + 0x80); |
2016 | 0 | op[1] = (BYTE)nbSeq; |
2017 | 0 | op+=2; |
2018 | 0 | } else { |
2019 | 0 | op[0]=0xFF; |
2020 | 0 | MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ)); |
2021 | 0 | op+=3; |
2022 | 0 | } |
2023 | 0 | assert(op <= oend); |
2024 | 0 | if (nbSeq==0) { |
2025 | | /* Copy the old tables over as if we repeated them */ |
2026 | 0 | memcpy(&nextEntropy->fse, &prevEntropy->fse, sizeof(prevEntropy->fse)); |
2027 | 0 | return (size_t)(op - ostart); |
2028 | 0 | } |
2029 | | |
2030 | | /* seqHead : flags for FSE encoding type */ |
2031 | 0 | seqHead = op++; |
2032 | 0 | assert(op <= oend); |
2033 | | |
2034 | | /* convert length/distances into codes */ |
2035 | 0 | ZSTD_seqToCodes(seqStorePtr); |
2036 | | /* build CTable for Literal Lengths */ |
2037 | 0 | { unsigned max = MaxLL; |
2038 | 0 | size_t const mostFrequent = HIST_countFast_wksp(count, &max, llCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */ |
2039 | 0 | DEBUGLOG(5, "Building LL table"); |
2040 | 0 | nextEntropy->fse.litlength_repeatMode = prevEntropy->fse.litlength_repeatMode; |
2041 | 0 | LLtype = ZSTD_selectEncodingType(&nextEntropy->fse.litlength_repeatMode, |
2042 | 0 | count, max, mostFrequent, nbSeq, |
2043 | 0 | LLFSELog, prevEntropy->fse.litlengthCTable, |
2044 | 0 | ZSTDInternalConstants::LL_defaultNorm, ZSTDInternalConstants::LL_defaultNormLog, |
2045 | 0 | ZSTD_defaultAllowed, strategy); |
2046 | 0 | assert(set_basic < set_compressed && set_rle < set_compressed); |
2047 | 0 | assert(!(LLtype < set_compressed && nextEntropy->fse.litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */ |
2048 | 0 | { size_t const countSize = ZSTD_buildCTable( |
2049 | 0 | op, (size_t)(oend - op), |
2050 | 0 | CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype, |
2051 | 0 | count, max, llCodeTable, nbSeq, |
2052 | 0 | ZSTDInternalConstants::LL_defaultNorm, ZSTDInternalConstants::LL_defaultNormLog, MaxLL, |
2053 | 0 | prevEntropy->fse.litlengthCTable, |
2054 | 0 | sizeof(prevEntropy->fse.litlengthCTable), |
2055 | 0 | entropyWorkspace, entropyWkspSize); |
2056 | 0 | FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed"); |
2057 | 0 | if (LLtype == set_compressed) |
2058 | 0 | lastNCount = op; |
2059 | 0 | op += countSize; |
2060 | 0 | assert(op <= oend); |
2061 | 0 | } } |
2062 | | /* build CTable for Offsets */ |
2063 | 0 | { unsigned max = MaxOff; |
2064 | 0 | size_t const mostFrequent = HIST_countFast_wksp( |
2065 | 0 | count, &max, ofCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */ |
2066 | | /* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */ |
2067 | 0 | ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed; |
2068 | 0 | DEBUGLOG(5, "Building OF table"); |
2069 | 0 | nextEntropy->fse.offcode_repeatMode = prevEntropy->fse.offcode_repeatMode; |
2070 | 0 | Offtype = ZSTD_selectEncodingType(&nextEntropy->fse.offcode_repeatMode, |
2071 | 0 | count, max, mostFrequent, nbSeq, |
2072 | 0 | OffFSELog, prevEntropy->fse.offcodeCTable, |
2073 | 0 | ZSTDInternalConstants::OF_defaultNorm, ZSTDInternalConstants::OF_defaultNormLog, |
2074 | 0 | defaultPolicy, strategy); |
2075 | 0 | assert(!(Offtype < set_compressed && nextEntropy->fse.offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */ |
2076 | 0 | { size_t const countSize = ZSTD_buildCTable( |
2077 | 0 | op, (size_t)(oend - op), |
2078 | 0 | CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype, |
2079 | 0 | count, max, ofCodeTable, nbSeq, |
2080 | 0 | ZSTDInternalConstants::OF_defaultNorm, ZSTDInternalConstants::OF_defaultNormLog, DefaultMaxOff, |
2081 | 0 | prevEntropy->fse.offcodeCTable, |
2082 | 0 | sizeof(prevEntropy->fse.offcodeCTable), |
2083 | 0 | entropyWorkspace, entropyWkspSize); |
2084 | 0 | FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed"); |
2085 | 0 | if (Offtype == set_compressed) |
2086 | 0 | lastNCount = op; |
2087 | 0 | op += countSize; |
2088 | 0 | assert(op <= oend); |
2089 | 0 | } } |
2090 | | /* build CTable for MatchLengths */ |
2091 | 0 | { unsigned max = MaxML; |
2092 | 0 | size_t const mostFrequent = HIST_countFast_wksp( |
2093 | 0 | count, &max, mlCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */ |
2094 | 0 | DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op)); |
2095 | 0 | nextEntropy->fse.matchlength_repeatMode = prevEntropy->fse.matchlength_repeatMode; |
2096 | 0 | MLtype = ZSTD_selectEncodingType(&nextEntropy->fse.matchlength_repeatMode, |
2097 | 0 | count, max, mostFrequent, nbSeq, |
2098 | 0 | MLFSELog, prevEntropy->fse.matchlengthCTable, |
2099 | 0 | ZSTDInternalConstants::ML_defaultNorm, ZSTDInternalConstants::ML_defaultNormLog, |
2100 | 0 | ZSTD_defaultAllowed, strategy); |
2101 | 0 | assert(!(MLtype < set_compressed && nextEntropy->fse.matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */ |
2102 | 0 | { size_t const countSize = ZSTD_buildCTable( |
2103 | 0 | op, (size_t)(oend - op), |
2104 | 0 | CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype, |
2105 | 0 | count, max, mlCodeTable, nbSeq, |
2106 | 0 | ZSTDInternalConstants::ML_defaultNorm, ZSTDInternalConstants::ML_defaultNormLog, MaxML, |
2107 | 0 | prevEntropy->fse.matchlengthCTable, |
2108 | 0 | sizeof(prevEntropy->fse.matchlengthCTable), |
2109 | 0 | entropyWorkspace, entropyWkspSize); |
2110 | 0 | FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed"); |
2111 | 0 | if (MLtype == set_compressed) |
2112 | 0 | lastNCount = op; |
2113 | 0 | op += countSize; |
2114 | 0 | assert(op <= oend); |
2115 | 0 | } } |
2116 | | |
2117 | 0 | *seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2)); |
2118 | |
|
2119 | 0 | { size_t const bitstreamSize = ZSTD_encodeSequences( |
2120 | 0 | op, (size_t)(oend - op), |
2121 | 0 | CTable_MatchLength, mlCodeTable, |
2122 | 0 | CTable_OffsetBits, ofCodeTable, |
2123 | 0 | CTable_LitLength, llCodeTable, |
2124 | 0 | sequences, nbSeq, |
2125 | 0 | longOffsets, bmi2); |
2126 | 0 | FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed"); |
2127 | 0 | op += bitstreamSize; |
2128 | 0 | assert(op <= oend); |
2129 | | /* zstd versions <= 1.3.4 mistakenly report corruption when |
2130 | | * FSE_readNCount() receives a buffer < 4 bytes. |
2131 | | * Fixed by https://github.com/facebook/zstd/pull/1146. |
2132 | | * This can happen when the last set_compressed table present is 2 |
2133 | | * bytes and the bitstream is only one byte. |
2134 | | * In this exceedingly rare case, we will simply emit an uncompressed |
2135 | | * block, since it isn't worth optimizing. |
2136 | | */ |
2137 | 0 | if (lastNCount && (op - lastNCount) < 4) { |
2138 | | /* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */ |
2139 | 0 | assert(op - lastNCount == 3); |
2140 | 0 | DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by " |
2141 | 0 | "emitting an uncompressed block."); |
2142 | 0 | return 0; |
2143 | 0 | } |
2144 | 0 | } |
2145 | | |
2146 | 0 | DEBUGLOG(5, "compressed block size : %u", (unsigned)(op - ostart)); |
2147 | 0 | return (size_t)(op - ostart); |
2148 | 0 | } |
2149 | | |
2150 | | MEM_STATIC size_t |
2151 | | ZSTD_compressSequences(seqStore_t* seqStorePtr, |
2152 | | const ZSTD_entropyCTables_t* prevEntropy, |
2153 | | ZSTD_entropyCTables_t* nextEntropy, |
2154 | | const ZSTD_CCtx_params* cctxParams, |
2155 | | void* dst, size_t dstCapacity, |
2156 | | size_t srcSize, |
2157 | | void* entropyWorkspace, size_t entropyWkspSize, |
2158 | | int bmi2) |
2159 | 0 | { |
2160 | 0 | size_t const cSize = ZSTD_compressSequences_internal( |
2161 | 0 | seqStorePtr, prevEntropy, nextEntropy, cctxParams, |
2162 | 0 | dst, dstCapacity, |
2163 | 0 | entropyWorkspace, entropyWkspSize, bmi2); |
2164 | 0 | if (cSize == 0) return 0; |
2165 | | /* When srcSize <= dstCapacity, there is enough space to write a raw uncompressed block. |
2166 | | * Since we ran out of space, block must be not compressible, so fall back to raw uncompressed block. |
2167 | | */ |
2168 | 0 | if ((cSize == ERROR(dstSize_tooSmall)) & (srcSize <= dstCapacity)) |
2169 | 0 | return 0; /* block not compressed */ |
2170 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressSequences_internal failed"); |
2171 | | |
2172 | | /* Check compressibility */ |
2173 | 0 | { size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, cctxParams->cParams.strategy); |
2174 | 0 | if (cSize >= maxCSize) return 0; /* block not compressed */ |
2175 | 0 | } |
2176 | | |
2177 | 0 | return cSize; |
2178 | 0 | } |
2179 | | |
2180 | | /* ZSTD_selectBlockCompressor() : |
2181 | | * Not static, but internal use only (used by long distance matcher) |
2182 | | * assumption : strat is a valid strategy */ |
2183 | | ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode) |
2184 | 0 | { |
2185 | 0 | static const ZSTD_blockCompressor blockCompressor[3][ZSTD_STRATEGY_MAX+1] = { |
2186 | 0 | { ZSTD_compressBlock_fast /* default for 0 */, |
2187 | 0 | ZSTD_compressBlock_fast, |
2188 | 0 | ZSTD_compressBlock_doubleFast, |
2189 | 0 | ZSTD_compressBlock_greedy, |
2190 | 0 | ZSTD_compressBlock_lazy, |
2191 | 0 | ZSTD_compressBlock_lazy2, |
2192 | 0 | ZSTD_compressBlock_btlazy2, |
2193 | 0 | ZSTD_compressBlock_btopt, |
2194 | 0 | ZSTD_compressBlock_btultra, |
2195 | 0 | ZSTD_compressBlock_btultra2 }, |
2196 | 0 | { ZSTD_compressBlock_fast_extDict /* default for 0 */, |
2197 | 0 | ZSTD_compressBlock_fast_extDict, |
2198 | 0 | ZSTD_compressBlock_doubleFast_extDict, |
2199 | 0 | ZSTD_compressBlock_greedy_extDict, |
2200 | 0 | ZSTD_compressBlock_lazy_extDict, |
2201 | 0 | ZSTD_compressBlock_lazy2_extDict, |
2202 | 0 | ZSTD_compressBlock_btlazy2_extDict, |
2203 | 0 | ZSTD_compressBlock_btopt_extDict, |
2204 | 0 | ZSTD_compressBlock_btultra_extDict, |
2205 | 0 | ZSTD_compressBlock_btultra_extDict }, |
2206 | 0 | { ZSTD_compressBlock_fast_dictMatchState /* default for 0 */, |
2207 | 0 | ZSTD_compressBlock_fast_dictMatchState, |
2208 | 0 | ZSTD_compressBlock_doubleFast_dictMatchState, |
2209 | 0 | ZSTD_compressBlock_greedy_dictMatchState, |
2210 | 0 | ZSTD_compressBlock_lazy_dictMatchState, |
2211 | 0 | ZSTD_compressBlock_lazy2_dictMatchState, |
2212 | 0 | ZSTD_compressBlock_btlazy2_dictMatchState, |
2213 | 0 | ZSTD_compressBlock_btopt_dictMatchState, |
2214 | 0 | ZSTD_compressBlock_btultra_dictMatchState, |
2215 | 0 | ZSTD_compressBlock_btultra_dictMatchState } |
2216 | 0 | }; |
2217 | 0 | ZSTD_blockCompressor selectedCompressor; |
2218 | 0 | ZSTD_STATIC_ASSERT((unsigned)ZSTD_fast == 1); |
2219 | |
|
2220 | 0 | assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat)); |
2221 | 0 | selectedCompressor = blockCompressor[(int)dictMode][(int)strat]; |
2222 | 0 | assert(selectedCompressor != NULL); |
2223 | 0 | return selectedCompressor; |
2224 | 0 | } |
2225 | | |
2226 | | static void ZSTD_storeLastLiterals(seqStore_t* seqStorePtr, |
2227 | | const BYTE* anchor, size_t lastLLSize) |
2228 | 0 | { |
2229 | 0 | memcpy(seqStorePtr->lit, anchor, lastLLSize); |
2230 | 0 | seqStorePtr->lit += lastLLSize; |
2231 | 0 | } |
2232 | | |
2233 | | void ZSTD_resetSeqStore(seqStore_t* ssPtr) |
2234 | 0 | { |
2235 | 0 | ssPtr->lit = ssPtr->litStart; |
2236 | 0 | ssPtr->sequences = ssPtr->sequencesStart; |
2237 | 0 | ssPtr->longLengthID = 0; |
2238 | 0 | } |
2239 | | |
2240 | | typedef enum { ZSTDbss_compress, ZSTDbss_noCompress } ZSTD_buildSeqStore_e; |
2241 | | |
2242 | | static size_t ZSTD_buildSeqStore(ZSTD_CCtx* zc, const void* src, size_t srcSize) |
2243 | 0 | { |
2244 | 0 | ZSTD_matchState_t* const ms = &zc->blockState.matchState; |
2245 | 0 | DEBUGLOG(5, "ZSTD_buildSeqStore (srcSize=%zu)", srcSize); |
2246 | 0 | assert(srcSize <= ZSTD_BLOCKSIZE_MAX); |
2247 | | /* Assert that we have correctly flushed the ctx params into the ms's copy */ |
2248 | 0 | ZSTD_assertEqualCParams(zc->appliedParams.cParams, ms->cParams); |
2249 | 0 | if (srcSize < MIN_CBLOCK_SIZE+ZSTDInternalConstants::ZSTD_blockHeaderSize+1) { |
2250 | 0 | ZSTD_ldm_skipSequences(&zc->externSeqStore, srcSize, zc->appliedParams.cParams.minMatch); |
2251 | 0 | return ZSTDbss_noCompress; /* don't even attempt compression below a certain srcSize */ |
2252 | 0 | } |
2253 | 0 | ZSTD_resetSeqStore(&(zc->seqStore)); |
2254 | | /* required for optimal parser to read stats from dictionary */ |
2255 | 0 | ms->opt.symbolCosts = &zc->blockState.prevCBlock->entropy; |
2256 | | /* tell the optimal parser how we expect to compress literals */ |
2257 | 0 | ms->opt.literalCompressionMode = zc->appliedParams.literalCompressionMode; |
2258 | | /* a gap between an attached dict and the current window is not safe, |
2259 | | * they must remain adjacent, |
2260 | | * and when that stops being the case, the dict must be unset */ |
2261 | 0 | assert(ms->dictMatchState == NULL || ms->loadedDictEnd == ms->window.dictLimit); |
2262 | | |
2263 | | /* limited update after a very long match */ |
2264 | 0 | { const BYTE* const base = ms->window.base; |
2265 | 0 | const BYTE* const istart = (const BYTE*)src; |
2266 | 0 | const U32 current = (U32)(istart-base); |
2267 | 0 | if (sizeof(ptrdiff_t)==8) assert(istart - base < (ptrdiff_t)(U32)(-1)); /* ensure no overflow */ |
2268 | 0 | if (current > ms->nextToUpdate + 384) |
2269 | 0 | ms->nextToUpdate = current - MIN(192, (U32)(current - ms->nextToUpdate - 384)); |
2270 | 0 | } |
2271 | | |
2272 | | /* select and store sequences */ |
2273 | 0 | { ZSTD_dictMode_e const dictMode = ZSTD_matchState_dictMode(ms); |
2274 | 0 | size_t lastLLSize; |
2275 | 0 | { int i; |
2276 | 0 | for (i = 0; i < ZSTD_REP_NUM; ++i) |
2277 | 0 | zc->blockState.nextCBlock->rep[i] = zc->blockState.prevCBlock->rep[i]; |
2278 | 0 | } |
2279 | 0 | if (zc->externSeqStore.pos < zc->externSeqStore.size) { |
2280 | 0 | assert(!zc->appliedParams.ldmParams.enableLdm); |
2281 | | /* Updates ldmSeqStore.pos */ |
2282 | 0 | lastLLSize = |
2283 | 0 | ZSTD_ldm_blockCompress(&zc->externSeqStore, |
2284 | 0 | ms, &zc->seqStore, |
2285 | 0 | zc->blockState.nextCBlock->rep, |
2286 | 0 | src, srcSize); |
2287 | 0 | assert(zc->externSeqStore.pos <= zc->externSeqStore.size); |
2288 | 0 | } else if (zc->appliedParams.ldmParams.enableLdm) { |
2289 | 0 | rawSeqStore_t ldmSeqStore = {NULL, 0, 0, 0}; |
2290 | |
|
2291 | 0 | ldmSeqStore.seq = zc->ldmSequences; |
2292 | 0 | ldmSeqStore.capacity = zc->maxNbLdmSequences; |
2293 | | /* Updates ldmSeqStore.size */ |
2294 | 0 | FORWARD_IF_ERROR(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore, |
2295 | 0 | &zc->appliedParams.ldmParams, |
2296 | 0 | src, srcSize), ""); |
2297 | | /* Updates ldmSeqStore.pos */ |
2298 | 0 | lastLLSize = |
2299 | 0 | ZSTD_ldm_blockCompress(&ldmSeqStore, |
2300 | 0 | ms, &zc->seqStore, |
2301 | 0 | zc->blockState.nextCBlock->rep, |
2302 | 0 | src, srcSize); |
2303 | 0 | assert(ldmSeqStore.pos == ldmSeqStore.size); |
2304 | 0 | } else { /* not long range mode */ |
2305 | 0 | ZSTD_blockCompressor const blockCompressor = ZSTD_selectBlockCompressor(zc->appliedParams.cParams.strategy, dictMode); |
2306 | 0 | lastLLSize = blockCompressor(ms, &zc->seqStore, zc->blockState.nextCBlock->rep, src, srcSize); |
2307 | 0 | } |
2308 | 0 | { const BYTE* const lastLiterals = (const BYTE*)src + srcSize - lastLLSize; |
2309 | 0 | ZSTD_storeLastLiterals(&zc->seqStore, lastLiterals, lastLLSize); |
2310 | 0 | } } |
2311 | 0 | return ZSTDbss_compress; |
2312 | 0 | } |
2313 | | |
2314 | | static void ZSTD_copyBlockSequences(ZSTD_CCtx* zc) |
2315 | 0 | { |
2316 | 0 | const seqStore_t* seqStore = ZSTD_getSeqStore(zc); |
2317 | 0 | const seqDef* seqs = seqStore->sequencesStart; |
2318 | 0 | size_t seqsSize = seqStore->sequences - seqs; |
2319 | |
|
2320 | 0 | ZSTD_Sequence* outSeqs = &zc->seqCollector.seqStart[zc->seqCollector.seqIndex]; |
2321 | 0 | size_t i; size_t position; int repIdx; |
2322 | |
|
2323 | 0 | assert(zc->seqCollector.seqIndex + 1 < zc->seqCollector.maxSequences); |
2324 | 0 | for (i = 0, position = 0; i < seqsSize; ++i) { |
2325 | 0 | outSeqs[i].offset = seqs[i].offset; |
2326 | 0 | outSeqs[i].litLength = seqs[i].litLength; |
2327 | 0 | outSeqs[i].matchLength = seqs[i].matchLength + MINMATCH; |
2328 | |
|
2329 | 0 | if (i == seqStore->longLengthPos) { |
2330 | 0 | if (seqStore->longLengthID == 1) { |
2331 | 0 | outSeqs[i].litLength += 0x10000; |
2332 | 0 | } else if (seqStore->longLengthID == 2) { |
2333 | 0 | outSeqs[i].matchLength += 0x10000; |
2334 | 0 | } |
2335 | 0 | } |
2336 | |
|
2337 | 0 | if (outSeqs[i].offset <= ZSTD_REP_NUM) { |
2338 | 0 | outSeqs[i].rep = outSeqs[i].offset; |
2339 | 0 | repIdx = (unsigned int)i - outSeqs[i].offset; |
2340 | |
|
2341 | 0 | if (outSeqs[i].litLength == 0) { |
2342 | 0 | if (outSeqs[i].offset < 3) { |
2343 | 0 | --repIdx; |
2344 | 0 | } else { |
2345 | 0 | repIdx = (unsigned int)i - 1; |
2346 | 0 | } |
2347 | 0 | ++outSeqs[i].rep; |
2348 | 0 | } |
2349 | 0 | assert(repIdx >= -3); |
2350 | 0 | outSeqs[i].offset = repIdx >= 0 ? outSeqs[repIdx].offset : ZSTDInternalConstants::repStartValue[-repIdx - 1]; |
2351 | 0 | if (outSeqs[i].rep == 4) { |
2352 | 0 | --outSeqs[i].offset; |
2353 | 0 | } |
2354 | 0 | } else { |
2355 | 0 | outSeqs[i].offset -= ZSTD_REP_NUM; |
2356 | 0 | } |
2357 | |
|
2358 | 0 | position += outSeqs[i].litLength; |
2359 | 0 | outSeqs[i].matchPos = (unsigned int)position; |
2360 | 0 | position += outSeqs[i].matchLength; |
2361 | 0 | } |
2362 | 0 | zc->seqCollector.seqIndex += seqsSize; |
2363 | 0 | } |
2364 | | |
2365 | | size_t ZSTD_getSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs, |
2366 | | size_t outSeqsSize, const void* src, size_t srcSize) |
2367 | 0 | { |
2368 | 0 | const size_t dstCapacity = ZSTD_compressBound(srcSize); |
2369 | 0 | void* dst = ZSTD_malloc(dstCapacity, ZSTDInternalConstants::ZSTD_defaultCMem); |
2370 | 0 | SeqCollector seqCollector; |
2371 | |
|
2372 | 0 | RETURN_ERROR_IF(dst == NULL, memory_allocation, "NULL pointer!"); |
2373 | |
|
2374 | 0 | seqCollector.collectSequences = 1; |
2375 | 0 | seqCollector.seqStart = outSeqs; |
2376 | 0 | seqCollector.seqIndex = 0; |
2377 | 0 | seqCollector.maxSequences = outSeqsSize; |
2378 | 0 | zc->seqCollector = seqCollector; |
2379 | |
|
2380 | 0 | ZSTD_compress2(zc, dst, dstCapacity, src, srcSize); |
2381 | 0 | ZSTD_free(dst, ZSTDInternalConstants::ZSTD_defaultCMem); |
2382 | 0 | return zc->seqCollector.seqIndex; |
2383 | 0 | } |
2384 | | |
2385 | | /* Returns true if the given block is a RLE block */ |
2386 | 0 | static int ZSTD_isRLE(const BYTE *ip, size_t length) { |
2387 | 0 | size_t i; |
2388 | 0 | if (length < 2) return 1; |
2389 | 0 | for (i = 1; i < length; ++i) { |
2390 | 0 | if (ip[0] != ip[i]) return 0; |
2391 | 0 | } |
2392 | 0 | return 1; |
2393 | 0 | } |
2394 | | |
2395 | | /* Returns true if the given block may be RLE. |
2396 | | * This is just a heuristic based on the compressibility. |
2397 | | * It may return both false positives and false negatives. |
2398 | | */ |
2399 | | static int ZSTD_maybeRLE(seqStore_t const* seqStore) |
2400 | 0 | { |
2401 | 0 | size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart); |
2402 | 0 | size_t const nbLits = (size_t)(seqStore->lit - seqStore->litStart); |
2403 | |
|
2404 | 0 | return nbSeqs < 4 && nbLits < 10; |
2405 | 0 | } |
2406 | | |
2407 | | static void ZSTD_confirmRepcodesAndEntropyTables(ZSTD_CCtx* zc) |
2408 | 0 | { |
2409 | 0 | ZSTD_compressedBlockState_t* const tmp = zc->blockState.prevCBlock; |
2410 | 0 | zc->blockState.prevCBlock = zc->blockState.nextCBlock; |
2411 | 0 | zc->blockState.nextCBlock = tmp; |
2412 | 0 | } |
2413 | | |
2414 | | static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc, |
2415 | | void* dst, size_t dstCapacity, |
2416 | | const void* src, size_t srcSize, U32 frame) |
2417 | 0 | { |
2418 | | /* This the upper bound for the length of an rle block. |
2419 | | * This isn't the actual upper bound. Finding the real threshold |
2420 | | * needs further investigation. |
2421 | | */ |
2422 | 0 | const U32 rleMaxLength = 25; |
2423 | 0 | size_t cSize; |
2424 | 0 | const BYTE* ip = (const BYTE*)src; |
2425 | 0 | BYTE* op = (BYTE*)dst; |
2426 | 0 | DEBUGLOG(5, "ZSTD_compressBlock_internal (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u)", |
2427 | 0 | (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit, |
2428 | 0 | (unsigned)zc->blockState.matchState.nextToUpdate); |
2429 | |
|
2430 | 0 | { const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize); |
2431 | 0 | FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed"); |
2432 | 0 | if (bss == ZSTDbss_noCompress) { cSize = 0; goto out; } |
2433 | 0 | } |
2434 | | |
2435 | 0 | if (zc->seqCollector.collectSequences) { |
2436 | 0 | ZSTD_copyBlockSequences(zc); |
2437 | 0 | return 0; |
2438 | 0 | } |
2439 | | |
2440 | | /* encode sequences and literals */ |
2441 | 0 | cSize = ZSTD_compressSequences(&zc->seqStore, |
2442 | 0 | &zc->blockState.prevCBlock->entropy, &zc->blockState.nextCBlock->entropy, |
2443 | 0 | &zc->appliedParams, |
2444 | 0 | dst, dstCapacity, |
2445 | 0 | srcSize, |
2446 | 0 | zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */, |
2447 | 0 | zc->bmi2); |
2448 | |
|
2449 | 0 | if (frame && |
2450 | | /* We don't want to emit our first block as a RLE even if it qualifies because |
2451 | | * doing so will cause the decoder (cli only) to throw a "should consume all input error." |
2452 | | * This is only an issue for zstd <= v1.4.3 |
2453 | | */ |
2454 | 0 | !zc->isFirstBlock && |
2455 | 0 | cSize < rleMaxLength && |
2456 | 0 | ZSTD_isRLE(ip, srcSize)) |
2457 | 0 | { |
2458 | 0 | cSize = 1; |
2459 | 0 | op[0] = ip[0]; |
2460 | 0 | } |
2461 | |
|
2462 | 0 | out: |
2463 | 0 | if (!ZSTD_isError(cSize) && cSize > 1) { |
2464 | 0 | ZSTD_confirmRepcodesAndEntropyTables(zc); |
2465 | 0 | } |
2466 | | /* We check that dictionaries have offset codes available for the first |
2467 | | * block. After the first block, the offcode table might not have large |
2468 | | * enough codes to represent the offsets in the data. |
2469 | | */ |
2470 | 0 | if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid) |
2471 | 0 | zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check; |
2472 | |
|
2473 | 0 | return cSize; |
2474 | 0 | } |
2475 | | |
2476 | | static size_t ZSTD_compressBlock_targetCBlockSize_body(ZSTD_CCtx* zc, |
2477 | | void* dst, size_t dstCapacity, |
2478 | | const void* src, size_t srcSize, |
2479 | | const size_t bss, U32 lastBlock) |
2480 | 0 | { |
2481 | 0 | DEBUGLOG(6, "Attempting ZSTD_compressSuperBlock()"); |
2482 | 0 | if (bss == ZSTDbss_compress) { |
2483 | 0 | if (/* We don't want to emit our first block as a RLE even if it qualifies because |
2484 | | * doing so will cause the decoder (cli only) to throw a "should consume all input error." |
2485 | | * This is only an issue for zstd <= v1.4.3 |
2486 | | */ |
2487 | 0 | !zc->isFirstBlock && |
2488 | 0 | ZSTD_maybeRLE(&zc->seqStore) && |
2489 | 0 | ZSTD_isRLE((BYTE const*)src, srcSize)) |
2490 | 0 | { |
2491 | 0 | return ZSTD_rleCompressBlock(dst, dstCapacity, *(BYTE const*)src, srcSize, lastBlock); |
2492 | 0 | } |
2493 | | /* Attempt superblock compression. |
2494 | | * |
2495 | | * Note that compressed size of ZSTD_compressSuperBlock() is not bound by the |
2496 | | * standard ZSTD_compressBound(). This is a problem, because even if we have |
2497 | | * space now, taking an extra byte now could cause us to run out of space later |
2498 | | * and violate ZSTD_compressBound(). |
2499 | | * |
2500 | | * Define blockBound(blockSize) = blockSize + ZSTD_blockHeaderSize. |
2501 | | * |
2502 | | * In order to respect ZSTD_compressBound() we must attempt to emit a raw |
2503 | | * uncompressed block in these cases: |
2504 | | * * cSize == 0: Return code for an uncompressed block. |
2505 | | * * cSize == dstSize_tooSmall: We may have expanded beyond blockBound(srcSize). |
2506 | | * ZSTD_noCompressBlock() will return dstSize_tooSmall if we are really out of |
2507 | | * output space. |
2508 | | * * cSize >= blockBound(srcSize): We have expanded the block too much so |
2509 | | * emit an uncompressed block. |
2510 | | */ |
2511 | 0 | { |
2512 | 0 | size_t const cSize = ZSTD_compressSuperBlock(zc, dst, dstCapacity, src, srcSize, lastBlock); |
2513 | 0 | if (cSize != ERROR(dstSize_tooSmall)) { |
2514 | 0 | size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, zc->appliedParams.cParams.strategy); |
2515 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressSuperBlock failed"); |
2516 | 0 | if (cSize != 0 && cSize < maxCSize + ZSTDInternalConstants::ZSTD_blockHeaderSize) { |
2517 | 0 | ZSTD_confirmRepcodesAndEntropyTables(zc); |
2518 | 0 | return cSize; |
2519 | 0 | } |
2520 | 0 | } |
2521 | 0 | } |
2522 | 0 | } |
2523 | | |
2524 | 0 | DEBUGLOG(6, "Resorting to ZSTD_noCompressBlock()"); |
2525 | | /* Superblock compression failed, attempt to emit a single no compress block. |
2526 | | * The decoder will be able to stream this block since it is uncompressed. |
2527 | | */ |
2528 | 0 | return ZSTD_noCompressBlock(dst, dstCapacity, src, srcSize, lastBlock); |
2529 | 0 | } |
2530 | | |
2531 | | static size_t ZSTD_compressBlock_targetCBlockSize(ZSTD_CCtx* zc, |
2532 | | void* dst, size_t dstCapacity, |
2533 | | const void* src, size_t srcSize, |
2534 | | U32 lastBlock) |
2535 | 0 | { |
2536 | 0 | size_t cSize = 0; |
2537 | 0 | const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize); |
2538 | 0 | DEBUGLOG(5, "ZSTD_compressBlock_targetCBlockSize (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u, srcSize=%zu)", |
2539 | 0 | (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit, (unsigned)zc->blockState.matchState.nextToUpdate, srcSize); |
2540 | 0 | FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed"); |
2541 | |
|
2542 | 0 | cSize = ZSTD_compressBlock_targetCBlockSize_body(zc, dst, dstCapacity, src, srcSize, bss, lastBlock); |
2543 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize_body failed"); |
2544 | |
|
2545 | 0 | if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid) |
2546 | 0 | zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check; |
2547 | |
|
2548 | 0 | return cSize; |
2549 | 0 | } |
2550 | | |
2551 | | static void ZSTD_overflowCorrectIfNeeded(ZSTD_matchState_t* ms, |
2552 | | ZSTD_cwksp* ws, |
2553 | | ZSTD_CCtx_params const* params, |
2554 | | void const* ip, |
2555 | | void const* iend) |
2556 | 0 | { |
2557 | 0 | if (ZSTD_window_needOverflowCorrection(ms->window, iend)) { |
2558 | 0 | U32 const maxDist = (U32)1 << params->cParams.windowLog; |
2559 | 0 | U32 const cycleLog = ZSTD_cycleLog(params->cParams.chainLog, params->cParams.strategy); |
2560 | 0 | U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, maxDist, ip); |
2561 | 0 | ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30); |
2562 | 0 | ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30); |
2563 | 0 | ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31); |
2564 | 0 | ZSTD_cwksp_mark_tables_dirty(ws); |
2565 | 0 | ZSTD_reduceIndex(ms, params, correction); |
2566 | 0 | ZSTD_cwksp_mark_tables_clean(ws); |
2567 | 0 | if (ms->nextToUpdate < correction) ms->nextToUpdate = 0; |
2568 | 0 | else ms->nextToUpdate -= correction; |
2569 | | /* invalidate dictionaries on overflow correction */ |
2570 | 0 | ms->loadedDictEnd = 0; |
2571 | 0 | ms->dictMatchState = NULL; |
2572 | 0 | } |
2573 | 0 | } |
2574 | | |
2575 | | /*! ZSTD_compress_frameChunk() : |
2576 | | * Compress a chunk of data into one or multiple blocks. |
2577 | | * All blocks will be terminated, all input will be consumed. |
2578 | | * Function will issue an error if there is not enough `dstCapacity` to hold the compressed content. |
2579 | | * Frame is supposed already started (header already produced) |
2580 | | * @return : compressed size, or an error code |
2581 | | */ |
2582 | | static size_t ZSTD_compress_frameChunk (ZSTD_CCtx* cctx, |
2583 | | void* dst, size_t dstCapacity, |
2584 | | const void* src, size_t srcSize, |
2585 | | U32 lastFrameChunk) |
2586 | 0 | { |
2587 | 0 | size_t blockSize = cctx->blockSize; |
2588 | 0 | size_t remaining = srcSize; |
2589 | 0 | const BYTE* ip = (const BYTE*)src; |
2590 | 0 | BYTE* const ostart = (BYTE*)dst; |
2591 | 0 | BYTE* op = ostart; |
2592 | 0 | U32 const maxDist = (U32)1 << cctx->appliedParams.cParams.windowLog; |
2593 | |
|
2594 | 0 | assert(cctx->appliedParams.cParams.windowLog <= ZSTD_WINDOWLOG_MAX); |
2595 | |
|
2596 | 0 | DEBUGLOG(5, "ZSTD_compress_frameChunk (blockSize=%u)", (unsigned)blockSize); |
2597 | 0 | if (cctx->appliedParams.fParams.checksumFlag && srcSize) |
2598 | 0 | XXH64_update(&cctx->xxhState, src, srcSize); |
2599 | |
|
2600 | 0 | while (remaining) { |
2601 | 0 | ZSTD_matchState_t* const ms = &cctx->blockState.matchState; |
2602 | 0 | U32 const lastBlock = lastFrameChunk & (blockSize >= remaining); |
2603 | |
|
2604 | 0 | RETURN_ERROR_IF(dstCapacity < ZSTDInternalConstants::ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE, |
2605 | 0 | dstSize_tooSmall, |
2606 | 0 | "not enough space to store compressed block"); |
2607 | 0 | if (remaining < blockSize) blockSize = remaining; |
2608 | |
|
2609 | 0 | ZSTD_overflowCorrectIfNeeded( |
2610 | 0 | ms, &cctx->workspace, &cctx->appliedParams, ip, ip + blockSize); |
2611 | 0 | ZSTD_checkDictValidity(&ms->window, ip + blockSize, maxDist, &ms->loadedDictEnd, &ms->dictMatchState); |
2612 | | |
2613 | | /* Ensure hash/chain table insertion resumes no sooner than lowlimit */ |
2614 | 0 | if (ms->nextToUpdate < ms->window.lowLimit) ms->nextToUpdate = ms->window.lowLimit; |
2615 | |
|
2616 | 0 | { size_t cSize; |
2617 | 0 | if (ZSTD_useTargetCBlockSize(&cctx->appliedParams)) { |
2618 | 0 | cSize = ZSTD_compressBlock_targetCBlockSize(cctx, op, dstCapacity, ip, blockSize, lastBlock); |
2619 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize failed"); |
2620 | 0 | assert(cSize > 0); |
2621 | 0 | assert(cSize <= blockSize + ZSTDInternalConstants::ZSTD_blockHeaderSize); |
2622 | 0 | } else { |
2623 | 0 | cSize = ZSTD_compressBlock_internal(cctx, |
2624 | 0 | op+ZSTDInternalConstants::ZSTD_blockHeaderSize, dstCapacity-ZSTDInternalConstants::ZSTD_blockHeaderSize, |
2625 | 0 | ip, blockSize, 1 /* frame */); |
2626 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_internal failed"); |
2627 | |
|
2628 | 0 | if (cSize == 0) { /* block is not compressible */ |
2629 | 0 | cSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock); |
2630 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed"); |
2631 | 0 | } else { |
2632 | 0 | U32 const cBlockHeader = cSize == 1 ? |
2633 | 0 | lastBlock + (((U32)bt_rle)<<1) + (U32)(blockSize << 3) : |
2634 | 0 | lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3); |
2635 | 0 | MEM_writeLE24(op, cBlockHeader); |
2636 | 0 | cSize += ZSTDInternalConstants::ZSTD_blockHeaderSize; |
2637 | 0 | } |
2638 | 0 | } |
2639 | | |
2640 | | |
2641 | 0 | ip += blockSize; |
2642 | 0 | assert(remaining >= blockSize); |
2643 | 0 | remaining -= blockSize; |
2644 | 0 | op += cSize; |
2645 | 0 | assert(dstCapacity >= cSize); |
2646 | 0 | dstCapacity -= cSize; |
2647 | 0 | cctx->isFirstBlock = 0; |
2648 | 0 | DEBUGLOG(5, "ZSTD_compress_frameChunk: adding a block of size %u", |
2649 | 0 | (unsigned)cSize); |
2650 | 0 | } } |
2651 | | |
2652 | 0 | if (lastFrameChunk && (op>ostart)) cctx->stage = ZSTDcs_ending; |
2653 | 0 | return (size_t)(op-ostart); |
2654 | 0 | } |
2655 | | |
2656 | | |
2657 | | static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity, |
2658 | | const ZSTD_CCtx_params* params, U64 pledgedSrcSize, U32 dictID) |
2659 | 0 | { BYTE* const op = (BYTE*)dst; |
2660 | 0 | U32 const dictIDSizeCodeLength = (dictID>0) + (dictID>=256) + (dictID>=65536); /* 0-3 */ |
2661 | 0 | U32 const dictIDSizeCode = params->fParams.noDictIDFlag ? 0 : dictIDSizeCodeLength; /* 0-3 */ |
2662 | 0 | U32 const checksumFlag = params->fParams.checksumFlag>0; |
2663 | 0 | U32 const windowSize = (U32)1 << params->cParams.windowLog; |
2664 | 0 | U32 const singleSegment = params->fParams.contentSizeFlag && (windowSize >= pledgedSrcSize); |
2665 | 0 | BYTE const windowLogByte = (BYTE)((params->cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3); |
2666 | 0 | U32 const fcsCode = params->fParams.contentSizeFlag ? |
2667 | 0 | (pledgedSrcSize>=256) + (pledgedSrcSize>=65536+256) + (pledgedSrcSize>=0xFFFFFFFFU) : 0; /* 0-3 */ |
2668 | 0 | BYTE const frameHeaderDescriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag<<2) + (singleSegment<<5) + (fcsCode<<6) ); |
2669 | 0 | size_t pos=0; |
2670 | |
|
2671 | 0 | assert(!(params->fParams.contentSizeFlag && pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)); |
2672 | 0 | RETURN_ERROR_IF(dstCapacity < ZSTD_FRAMEHEADERSIZE_MAX, dstSize_tooSmall, |
2673 | 0 | "dst buf is too small to fit worst-case frame header size."); |
2674 | 0 | DEBUGLOG(4, "ZSTD_writeFrameHeader : dictIDFlag : %u ; dictID : %u ; dictIDSizeCode : %u", |
2675 | 0 | !params->fParams.noDictIDFlag, (unsigned)dictID, (unsigned)dictIDSizeCode); |
2676 | |
|
2677 | 0 | if (params->format == ZSTD_f_zstd1) { |
2678 | 0 | MEM_writeLE32(dst, ZSTD_MAGICNUMBER); |
2679 | 0 | pos = 4; |
2680 | 0 | } |
2681 | 0 | op[pos++] = frameHeaderDescriptionByte; |
2682 | 0 | if (!singleSegment) op[pos++] = windowLogByte; |
2683 | 0 | switch(dictIDSizeCode) |
2684 | 0 | { |
2685 | 0 | default: assert(0); /* impossible */ |
2686 | 0 | case 0 : break; |
2687 | 0 | case 1 : op[pos] = (BYTE)(dictID); pos++; break; |
2688 | 0 | case 2 : MEM_writeLE16(op+pos, (U16)dictID); pos+=2; break; |
2689 | 0 | case 3 : MEM_writeLE32(op+pos, dictID); pos+=4; break; |
2690 | 0 | } |
2691 | 0 | switch(fcsCode) |
2692 | 0 | { |
2693 | 0 | default: assert(0); /* impossible */ |
2694 | 0 | case 0 : if (singleSegment) op[pos++] = (BYTE)(pledgedSrcSize); break; |
2695 | 0 | case 1 : MEM_writeLE16(op+pos, (U16)(pledgedSrcSize-256)); pos+=2; break; |
2696 | 0 | case 2 : MEM_writeLE32(op+pos, (U32)(pledgedSrcSize)); pos+=4; break; |
2697 | 0 | case 3 : MEM_writeLE64(op+pos, (U64)(pledgedSrcSize)); pos+=8; break; |
2698 | 0 | } |
2699 | 0 | return pos; |
2700 | 0 | } |
2701 | | |
2702 | | /* ZSTD_writeLastEmptyBlock() : |
2703 | | * output an empty Block with end-of-frame mark to complete a frame |
2704 | | * @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h)) |
2705 | | * or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize) |
2706 | | */ |
2707 | | size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity) |
2708 | 0 | { |
2709 | 0 | RETURN_ERROR_IF(dstCapacity < ZSTDInternalConstants::ZSTD_blockHeaderSize, dstSize_tooSmall, |
2710 | 0 | "dst buf is too small to write frame trailer empty block."); |
2711 | 0 | { U32 const cBlockHeader24 = 1 /*lastBlock*/ + (((U32)bt_raw)<<1); /* 0 size */ |
2712 | 0 | MEM_writeLE24(dst, cBlockHeader24); |
2713 | 0 | return ZSTDInternalConstants::ZSTD_blockHeaderSize; |
2714 | 0 | } |
2715 | 0 | } |
2716 | | |
2717 | | size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq) |
2718 | 0 | { |
2719 | 0 | RETURN_ERROR_IF(cctx->stage != ZSTDcs_init, stage_wrong, |
2720 | 0 | "wrong cctx stage"); |
2721 | 0 | RETURN_ERROR_IF(cctx->appliedParams.ldmParams.enableLdm, |
2722 | 0 | parameter_unsupported, |
2723 | 0 | "incompatible with ldm"); |
2724 | 0 | cctx->externSeqStore.seq = seq; |
2725 | 0 | cctx->externSeqStore.size = nbSeq; |
2726 | 0 | cctx->externSeqStore.capacity = nbSeq; |
2727 | 0 | cctx->externSeqStore.pos = 0; |
2728 | 0 | return 0; |
2729 | 0 | } |
2730 | | |
2731 | | |
2732 | | static size_t ZSTD_compressContinue_internal (ZSTD_CCtx* cctx, |
2733 | | void* dst, size_t dstCapacity, |
2734 | | const void* src, size_t srcSize, |
2735 | | U32 frame, U32 lastFrameChunk) |
2736 | 0 | { |
2737 | 0 | ZSTD_matchState_t* const ms = &cctx->blockState.matchState; |
2738 | 0 | size_t fhSize = 0; |
2739 | |
|
2740 | 0 | DEBUGLOG(5, "ZSTD_compressContinue_internal, stage: %u, srcSize: %u", |
2741 | 0 | cctx->stage, (unsigned)srcSize); |
2742 | 0 | RETURN_ERROR_IF(cctx->stage==ZSTDcs_created, stage_wrong, |
2743 | 0 | "missing init (ZSTD_compressBegin)"); |
2744 | |
|
2745 | 0 | if (frame && (cctx->stage==ZSTDcs_init)) { |
2746 | 0 | fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams, |
2747 | 0 | cctx->pledgedSrcSizePlusOne-1, cctx->dictID); |
2748 | 0 | FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed"); |
2749 | 0 | assert(fhSize <= dstCapacity); |
2750 | 0 | dstCapacity -= fhSize; |
2751 | 0 | dst = (char*)dst + fhSize; |
2752 | 0 | cctx->stage = ZSTDcs_ongoing; |
2753 | 0 | } |
2754 | | |
2755 | 0 | if (!srcSize) return fhSize; /* do not generate an empty block if no input */ |
2756 | | |
2757 | 0 | if (!ZSTD_window_update(&ms->window, src, srcSize)) { |
2758 | 0 | ms->nextToUpdate = ms->window.dictLimit; |
2759 | 0 | } |
2760 | 0 | if (cctx->appliedParams.ldmParams.enableLdm) { |
2761 | 0 | ZSTD_window_update(&cctx->ldmState.window, src, srcSize); |
2762 | 0 | } |
2763 | |
|
2764 | 0 | if (!frame) { |
2765 | | /* overflow check and correction for block mode */ |
2766 | 0 | ZSTD_overflowCorrectIfNeeded( |
2767 | 0 | ms, &cctx->workspace, &cctx->appliedParams, |
2768 | 0 | src, (BYTE const*)src + srcSize); |
2769 | 0 | } |
2770 | |
|
2771 | 0 | DEBUGLOG(5, "ZSTD_compressContinue_internal (blockSize=%u)", (unsigned)cctx->blockSize); |
2772 | 0 | { size_t const cSize = frame ? |
2773 | 0 | ZSTD_compress_frameChunk (cctx, dst, dstCapacity, src, srcSize, lastFrameChunk) : |
2774 | 0 | ZSTD_compressBlock_internal (cctx, dst, dstCapacity, src, srcSize, 0 /* frame */); |
2775 | 0 | FORWARD_IF_ERROR(cSize, "%s", frame ? "ZSTD_compress_frameChunk failed" : "ZSTD_compressBlock_internal failed"); |
2776 | 0 | cctx->consumedSrcSize += srcSize; |
2777 | 0 | cctx->producedCSize += (cSize + fhSize); |
2778 | 0 | assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0)); |
2779 | 0 | if (cctx->pledgedSrcSizePlusOne != 0) { /* control src size */ |
2780 | 0 | ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1); |
2781 | 0 | RETURN_ERROR_IF( |
2782 | 0 | cctx->consumedSrcSize+1 > cctx->pledgedSrcSizePlusOne, |
2783 | 0 | srcSize_wrong, |
2784 | 0 | "error : pledgedSrcSize = %u, while realSrcSize >= %u", |
2785 | 0 | (unsigned)cctx->pledgedSrcSizePlusOne-1, |
2786 | 0 | (unsigned)cctx->consumedSrcSize); |
2787 | 0 | } |
2788 | 0 | return cSize + fhSize; |
2789 | 0 | } |
2790 | 0 | } |
2791 | | |
2792 | | size_t ZSTD_compressContinue (ZSTD_CCtx* cctx, |
2793 | | void* dst, size_t dstCapacity, |
2794 | | const void* src, size_t srcSize) |
2795 | 0 | { |
2796 | 0 | DEBUGLOG(5, "ZSTD_compressContinue (srcSize=%u)", (unsigned)srcSize); |
2797 | 0 | return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1 /* frame mode */, 0 /* last chunk */); |
2798 | 0 | } |
2799 | | |
2800 | | |
2801 | | size_t ZSTD_getBlockSize(const ZSTD_CCtx* cctx) |
2802 | 0 | { |
2803 | 0 | ZSTD_compressionParameters const cParams = cctx->appliedParams.cParams; |
2804 | 0 | assert(!ZSTD_checkCParams(cParams)); |
2805 | 0 | return MIN (ZSTD_BLOCKSIZE_MAX, (U32)1 << cParams.windowLog); |
2806 | 0 | } |
2807 | | |
2808 | | size_t ZSTD_compressBlock(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize) |
2809 | 0 | { |
2810 | 0 | DEBUGLOG(5, "ZSTD_compressBlock: srcSize = %u", (unsigned)srcSize); |
2811 | 0 | { size_t const blockSizeMax = ZSTD_getBlockSize(cctx); |
2812 | 0 | RETURN_ERROR_IF(srcSize > blockSizeMax, srcSize_wrong, "input is larger than a block"); } |
2813 | | |
2814 | 0 | return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0 /* frame mode */, 0 /* last chunk */); |
2815 | 0 | } |
2816 | | |
2817 | | /*! ZSTD_loadDictionaryContent() : |
2818 | | * @return : 0, or an error code |
2819 | | */ |
2820 | | static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms, |
2821 | | ldmState_t* ls, |
2822 | | ZSTD_cwksp* ws, |
2823 | | ZSTD_CCtx_params const* params, |
2824 | | const void* src, size_t srcSize, |
2825 | | ZSTD_dictTableLoadMethod_e dtlm) |
2826 | 0 | { |
2827 | 0 | const BYTE* ip = (const BYTE*) src; |
2828 | 0 | const BYTE* const iend = ip + srcSize; |
2829 | |
|
2830 | 0 | ZSTD_window_update(&ms->window, src, srcSize); |
2831 | 0 | ms->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ms->window.base); |
2832 | |
|
2833 | 0 | if (params->ldmParams.enableLdm && ls != NULL) { |
2834 | 0 | ZSTD_window_update(&ls->window, src, srcSize); |
2835 | 0 | ls->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ls->window.base); |
2836 | 0 | } |
2837 | | |
2838 | | /* Assert that we the ms params match the params we're being given */ |
2839 | 0 | ZSTD_assertEqualCParams(params->cParams, ms->cParams); |
2840 | |
|
2841 | 0 | if (srcSize <= HASH_READ_SIZE) return 0; |
2842 | | |
2843 | 0 | while (iend - ip > HASH_READ_SIZE) { |
2844 | 0 | size_t const remaining = (size_t)(iend - ip); |
2845 | 0 | size_t const chunk = MIN(remaining, ZSTD_CHUNKSIZE_MAX); |
2846 | 0 | const BYTE* const ichunk = ip + chunk; |
2847 | |
|
2848 | 0 | ZSTD_overflowCorrectIfNeeded(ms, ws, params, ip, ichunk); |
2849 | |
|
2850 | 0 | if (params->ldmParams.enableLdm && ls != NULL) |
2851 | 0 | ZSTD_ldm_fillHashTable(ls, (const BYTE*)src, (const BYTE*)src + srcSize, ¶ms->ldmParams); |
2852 | |
|
2853 | 0 | switch(params->cParams.strategy) |
2854 | 0 | { |
2855 | 0 | case ZSTD_fast: |
2856 | 0 | ZSTD_fillHashTable(ms, ichunk, dtlm); |
2857 | 0 | break; |
2858 | 0 | case ZSTD_dfast: |
2859 | 0 | ZSTD_fillDoubleHashTable(ms, ichunk, dtlm); |
2860 | 0 | break; |
2861 | | |
2862 | 0 | case ZSTD_greedy: |
2863 | 0 | case ZSTD_lazy: |
2864 | 0 | case ZSTD_lazy2: |
2865 | 0 | if (chunk >= HASH_READ_SIZE) |
2866 | 0 | ZSTD_insertAndFindFirstIndex(ms, ichunk-HASH_READ_SIZE); |
2867 | 0 | break; |
2868 | | |
2869 | 0 | case ZSTD_btlazy2: /* we want the dictionary table fully sorted */ |
2870 | 0 | case ZSTD_btopt: |
2871 | 0 | case ZSTD_btultra: |
2872 | 0 | case ZSTD_btultra2: |
2873 | 0 | if (chunk >= HASH_READ_SIZE) |
2874 | 0 | ZSTD_updateTree(ms, ichunk-HASH_READ_SIZE, ichunk); |
2875 | 0 | break; |
2876 | | |
2877 | 0 | default: |
2878 | 0 | assert(0); /* not possible : not a valid strategy id */ |
2879 | 0 | } |
2880 | | |
2881 | 0 | ip = ichunk; |
2882 | 0 | } |
2883 | | |
2884 | 0 | ms->nextToUpdate = (U32)(iend - ms->window.base); |
2885 | 0 | return 0; |
2886 | 0 | } |
2887 | | |
2888 | | |
2889 | | /* Dictionaries that assign zero probability to symbols that show up causes problems |
2890 | | when FSE encoding. Refuse dictionaries that assign zero probability to symbols |
2891 | | that we may encounter during compression. |
2892 | | NOTE: This behavior is not standard and could be improved in the future. */ |
2893 | 0 | static size_t ZSTD_checkDictNCount(short* normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue) { |
2894 | 0 | U32 s; |
2895 | 0 | RETURN_ERROR_IF(dictMaxSymbolValue < maxSymbolValue, dictionary_corrupted, "dict fse tables don't have all symbols"); |
2896 | 0 | for (s = 0; s <= maxSymbolValue; ++s) { |
2897 | 0 | RETURN_ERROR_IF(normalizedCounter[s] == 0, dictionary_corrupted, "dict fse tables don't have all symbols"); |
2898 | 0 | } |
2899 | 0 | return 0; |
2900 | 0 | } |
2901 | | |
2902 | | size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace, |
2903 | | short* offcodeNCount, unsigned* offcodeMaxValue, |
2904 | | const void* const dict, size_t dictSize) |
2905 | 0 | { |
2906 | 0 | const BYTE* dictPtr = (const BYTE*)dict; /* skip magic num and dict ID */ |
2907 | 0 | const BYTE* const dictEnd = dictPtr + dictSize; |
2908 | 0 | dictPtr += 8; |
2909 | 0 | bs->entropy.huf.repeatMode = HUF_repeat_check; |
2910 | |
|
2911 | 0 | { unsigned maxSymbolValue = 255; |
2912 | 0 | unsigned hasZeroWeights = 1; |
2913 | 0 | size_t const hufHeaderSize = HUF_readCTable((HUF_CElt*)bs->entropy.huf.CTable, &maxSymbolValue, dictPtr, |
2914 | 0 | dictEnd-dictPtr, &hasZeroWeights); |
2915 | | |
2916 | | /* We only set the loaded table as valid if it contains all non-zero |
2917 | | * weights. Otherwise, we set it to check */ |
2918 | 0 | if (!hasZeroWeights) |
2919 | 0 | bs->entropy.huf.repeatMode = HUF_repeat_valid; |
2920 | |
|
2921 | 0 | RETURN_ERROR_IF(HUF_isError(hufHeaderSize), dictionary_corrupted, ""); |
2922 | 0 | RETURN_ERROR_IF(maxSymbolValue < 255, dictionary_corrupted, ""); |
2923 | 0 | dictPtr += hufHeaderSize; |
2924 | 0 | } |
2925 | | |
2926 | 0 | { unsigned offcodeLog; |
2927 | 0 | size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr); |
2928 | 0 | RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, ""); |
2929 | 0 | RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, ""); |
2930 | | /* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */ |
2931 | | /* fill all offset symbols to avoid garbage at end of table */ |
2932 | 0 | RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp( |
2933 | 0 | bs->entropy.fse.offcodeCTable, |
2934 | 0 | offcodeNCount, MaxOff, offcodeLog, |
2935 | 0 | workspace, HUF_WORKSPACE_SIZE)), |
2936 | 0 | dictionary_corrupted, ""); |
2937 | 0 | dictPtr += offcodeHeaderSize; |
2938 | 0 | } |
2939 | | |
2940 | 0 | { short matchlengthNCount[MaxML+1]; |
2941 | 0 | unsigned matchlengthMaxValue = MaxML, matchlengthLog; |
2942 | 0 | size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd-dictPtr); |
2943 | 0 | RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, ""); |
2944 | 0 | RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, ""); |
2945 | | /* Every match length code must have non-zero probability */ |
2946 | 0 | FORWARD_IF_ERROR( ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML), ""); |
2947 | 0 | RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp( |
2948 | 0 | bs->entropy.fse.matchlengthCTable, |
2949 | 0 | matchlengthNCount, matchlengthMaxValue, matchlengthLog, |
2950 | 0 | workspace, HUF_WORKSPACE_SIZE)), |
2951 | 0 | dictionary_corrupted, ""); |
2952 | 0 | dictPtr += matchlengthHeaderSize; |
2953 | 0 | } |
2954 | | |
2955 | 0 | { short litlengthNCount[MaxLL+1]; |
2956 | 0 | unsigned litlengthMaxValue = MaxLL, litlengthLog; |
2957 | 0 | size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd-dictPtr); |
2958 | 0 | RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, ""); |
2959 | 0 | RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, ""); |
2960 | | /* Every literal length code must have non-zero probability */ |
2961 | 0 | FORWARD_IF_ERROR( ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL), ""); |
2962 | 0 | RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp( |
2963 | 0 | bs->entropy.fse.litlengthCTable, |
2964 | 0 | litlengthNCount, litlengthMaxValue, litlengthLog, |
2965 | 0 | workspace, HUF_WORKSPACE_SIZE)), |
2966 | 0 | dictionary_corrupted, ""); |
2967 | 0 | dictPtr += litlengthHeaderSize; |
2968 | 0 | } |
2969 | | |
2970 | 0 | RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, ""); |
2971 | 0 | bs->rep[0] = MEM_readLE32(dictPtr+0); |
2972 | 0 | bs->rep[1] = MEM_readLE32(dictPtr+4); |
2973 | 0 | bs->rep[2] = MEM_readLE32(dictPtr+8); |
2974 | 0 | dictPtr += 12; |
2975 | |
|
2976 | 0 | return dictPtr - (const BYTE*)dict; |
2977 | 0 | } |
2978 | | |
2979 | | /* Dictionary format : |
2980 | | * See : |
2981 | | * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format |
2982 | | */ |
2983 | | /*! ZSTD_loadZstdDictionary() : |
2984 | | * @return : dictID, or an error code |
2985 | | * assumptions : magic number supposed already checked |
2986 | | * dictSize supposed >= 8 |
2987 | | */ |
2988 | | static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs, |
2989 | | ZSTD_matchState_t* ms, |
2990 | | ZSTD_cwksp* ws, |
2991 | | ZSTD_CCtx_params const* params, |
2992 | | const void* dict, size_t dictSize, |
2993 | | ZSTD_dictTableLoadMethod_e dtlm, |
2994 | | void* workspace) |
2995 | 0 | { |
2996 | 0 | const BYTE* dictPtr = (const BYTE*)dict; |
2997 | 0 | const BYTE* const dictEnd = dictPtr + dictSize; |
2998 | 0 | short offcodeNCount[MaxOff+1]; |
2999 | 0 | unsigned offcodeMaxValue = MaxOff; |
3000 | 0 | size_t dictID; |
3001 | 0 | size_t eSize; |
3002 | |
|
3003 | 0 | ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog))); |
3004 | 0 | assert(dictSize >= 8); |
3005 | 0 | assert(MEM_readLE32(dictPtr) == ZSTD_MAGIC_DICTIONARY); |
3006 | |
|
3007 | 0 | dictID = params->fParams.noDictIDFlag ? 0 : MEM_readLE32(dictPtr + 4 /* skip magic number */ ); |
3008 | 0 | eSize = ZSTD_loadCEntropy(bs, workspace, offcodeNCount, &offcodeMaxValue, dict, dictSize); |
3009 | 0 | FORWARD_IF_ERROR(eSize, "ZSTD_loadCEntropy failed"); |
3010 | 0 | dictPtr += eSize; |
3011 | |
|
3012 | 0 | { size_t const dictContentSize = (size_t)(dictEnd - dictPtr); |
3013 | 0 | U32 offcodeMax = MaxOff; |
3014 | 0 | if (dictContentSize <= ((U32)-1) - 128 KB) { |
3015 | 0 | U32 const maxOffset = (U32)dictContentSize + 128 KB; /* The maximum offset that must be supported */ |
3016 | 0 | offcodeMax = ZSTD_highbit32(maxOffset); /* Calculate minimum offset code required to represent maxOffset */ |
3017 | 0 | } |
3018 | | /* All offset values <= dictContentSize + 128 KB must be representable */ |
3019 | 0 | FORWARD_IF_ERROR(ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)), ""); |
3020 | | /* All repCodes must be <= dictContentSize and != 0*/ |
3021 | 0 | { U32 u; |
3022 | 0 | for (u=0; u<3; u++) { |
3023 | 0 | RETURN_ERROR_IF(bs->rep[u] == 0, dictionary_corrupted, ""); |
3024 | 0 | RETURN_ERROR_IF(bs->rep[u] > dictContentSize, dictionary_corrupted, ""); |
3025 | 0 | } } |
3026 | | |
3027 | 0 | bs->entropy.fse.offcode_repeatMode = FSE_repeat_valid; |
3028 | 0 | bs->entropy.fse.matchlength_repeatMode = FSE_repeat_valid; |
3029 | 0 | bs->entropy.fse.litlength_repeatMode = FSE_repeat_valid; |
3030 | 0 | FORWARD_IF_ERROR(ZSTD_loadDictionaryContent( |
3031 | 0 | ms, NULL, ws, params, dictPtr, dictContentSize, dtlm), ""); |
3032 | 0 | return dictID; |
3033 | 0 | } |
3034 | 0 | } |
3035 | | |
3036 | | /** ZSTD_compress_insertDictionary() : |
3037 | | * @return : dictID, or an error code */ |
3038 | | static size_t |
3039 | | ZSTD_compress_insertDictionary(ZSTD_compressedBlockState_t* bs, |
3040 | | ZSTD_matchState_t* ms, |
3041 | | ldmState_t* ls, |
3042 | | ZSTD_cwksp* ws, |
3043 | | const ZSTD_CCtx_params* params, |
3044 | | const void* dict, size_t dictSize, |
3045 | | ZSTD_dictContentType_e dictContentType, |
3046 | | ZSTD_dictTableLoadMethod_e dtlm, |
3047 | | void* workspace) |
3048 | 0 | { |
3049 | 0 | DEBUGLOG(4, "ZSTD_compress_insertDictionary (dictSize=%u)", (U32)dictSize); |
3050 | 0 | if ((dict==NULL) || (dictSize<8)) { |
3051 | 0 | RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, ""); |
3052 | 0 | return 0; |
3053 | 0 | } |
3054 | | |
3055 | 0 | ZSTD_reset_compressedBlockState(bs); |
3056 | | |
3057 | | /* dict restricted modes */ |
3058 | 0 | if (dictContentType == ZSTD_dct_rawContent) |
3059 | 0 | return ZSTD_loadDictionaryContent(ms, ls, ws, params, dict, dictSize, dtlm); |
3060 | | |
3061 | 0 | if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) { |
3062 | 0 | if (dictContentType == ZSTD_dct_auto) { |
3063 | 0 | DEBUGLOG(4, "raw content dictionary detected"); |
3064 | 0 | return ZSTD_loadDictionaryContent( |
3065 | 0 | ms, ls, ws, params, dict, dictSize, dtlm); |
3066 | 0 | } |
3067 | 0 | RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, ""); |
3068 | 0 | assert(0); /* impossible */ |
3069 | 0 | } |
3070 | | |
3071 | | /* dict as full zstd dictionary */ |
3072 | 0 | return ZSTD_loadZstdDictionary( |
3073 | 0 | bs, ms, ws, params, dict, dictSize, dtlm, workspace); |
3074 | 0 | } |
3075 | | |
3076 | 0 | #define ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF (128 KB) |
3077 | 0 | #define ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER (6) |
3078 | | |
3079 | | /*! ZSTD_compressBegin_internal() : |
3080 | | * @return : 0, or an error code */ |
3081 | | static size_t ZSTD_compressBegin_internal(ZSTD_CCtx* cctx, |
3082 | | const void* dict, size_t dictSize, |
3083 | | ZSTD_dictContentType_e dictContentType, |
3084 | | ZSTD_dictTableLoadMethod_e dtlm, |
3085 | | const ZSTD_CDict* cdict, |
3086 | | const ZSTD_CCtx_params* params, U64 pledgedSrcSize, |
3087 | | ZSTD_buffered_policy_e zbuff) |
3088 | 0 | { |
3089 | 0 | DEBUGLOG(4, "ZSTD_compressBegin_internal: wlog=%u", params->cParams.windowLog); |
3090 | | /* params are supposed to be fully validated at this point */ |
3091 | 0 | assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams))); |
3092 | 0 | assert(!((dict) && (cdict))); /* either dict or cdict, not both */ |
3093 | 0 | if ( (cdict) |
3094 | 0 | && (cdict->dictContentSize > 0) |
3095 | 0 | && ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF |
3096 | 0 | || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER |
3097 | 0 | || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN |
3098 | 0 | || cdict->compressionLevel == 0) |
3099 | 0 | && (params->attachDictPref != ZSTD_dictForceLoad) ) { |
3100 | 0 | return ZSTD_resetCCtx_usingCDict(cctx, cdict, params, pledgedSrcSize, zbuff); |
3101 | 0 | } |
3102 | | |
3103 | 0 | FORWARD_IF_ERROR( ZSTD_resetCCtx_internal(cctx, *params, pledgedSrcSize, |
3104 | 0 | ZSTDcrp_makeClean, zbuff) , ""); |
3105 | 0 | { size_t const dictID = cdict ? |
3106 | 0 | ZSTD_compress_insertDictionary( |
3107 | 0 | cctx->blockState.prevCBlock, &cctx->blockState.matchState, |
3108 | 0 | &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, cdict->dictContent, |
3109 | 0 | cdict->dictContentSize, dictContentType, dtlm, |
3110 | 0 | cctx->entropyWorkspace) |
3111 | 0 | : ZSTD_compress_insertDictionary( |
3112 | 0 | cctx->blockState.prevCBlock, &cctx->blockState.matchState, |
3113 | 0 | &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, dict, dictSize, |
3114 | 0 | dictContentType, dtlm, cctx->entropyWorkspace); |
3115 | 0 | FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed"); |
3116 | 0 | assert(dictID <= UINT_MAX); |
3117 | 0 | cctx->dictID = (U32)dictID; |
3118 | 0 | } |
3119 | 0 | return 0; |
3120 | 0 | } |
3121 | | |
3122 | | size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx, |
3123 | | const void* dict, size_t dictSize, |
3124 | | ZSTD_dictContentType_e dictContentType, |
3125 | | ZSTD_dictTableLoadMethod_e dtlm, |
3126 | | const ZSTD_CDict* cdict, |
3127 | | const ZSTD_CCtx_params* params, |
3128 | | unsigned long long pledgedSrcSize) |
3129 | 0 | { |
3130 | 0 | DEBUGLOG(4, "ZSTD_compressBegin_advanced_internal: wlog=%u", params->cParams.windowLog); |
3131 | | /* compression parameters verification and optimization */ |
3132 | 0 | FORWARD_IF_ERROR( ZSTD_checkCParams(params->cParams) , ""); |
3133 | 0 | return ZSTD_compressBegin_internal(cctx, |
3134 | 0 | dict, dictSize, dictContentType, dtlm, |
3135 | 0 | cdict, |
3136 | 0 | params, pledgedSrcSize, |
3137 | 0 | ZSTDb_not_buffered); |
3138 | 0 | } |
3139 | | |
3140 | | /*! ZSTD_compressBegin_advanced() : |
3141 | | * @return : 0, or an error code */ |
3142 | | size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, |
3143 | | const void* dict, size_t dictSize, |
3144 | | ZSTD_parameters params, unsigned long long pledgedSrcSize) |
3145 | 0 | { |
3146 | 0 | ZSTD_CCtx_params const cctxParams = |
3147 | 0 | ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, ¶ms); |
3148 | 0 | return ZSTD_compressBegin_advanced_internal(cctx, |
3149 | 0 | dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, |
3150 | 0 | NULL /*cdict*/, |
3151 | 0 | &cctxParams, pledgedSrcSize); |
3152 | 0 | } |
3153 | | |
3154 | | size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel) |
3155 | 0 | { |
3156 | 0 | ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize); |
3157 | 0 | ZSTD_CCtx_params const cctxParams = |
3158 | 0 | ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, ¶ms); |
3159 | 0 | DEBUGLOG(4, "ZSTD_compressBegin_usingDict (dictSize=%u)", (unsigned)dictSize); |
3160 | 0 | return ZSTD_compressBegin_internal(cctx, dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL, |
3161 | 0 | &cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, ZSTDb_not_buffered); |
3162 | 0 | } |
3163 | | |
3164 | | size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel) |
3165 | 0 | { |
3166 | 0 | return ZSTD_compressBegin_usingDict(cctx, NULL, 0, compressionLevel); |
3167 | 0 | } |
3168 | | |
3169 | | |
3170 | | /*! ZSTD_writeEpilogue() : |
3171 | | * Ends a frame. |
3172 | | * @return : nb of bytes written into dst (or an error code) */ |
3173 | | static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity) |
3174 | 0 | { |
3175 | 0 | BYTE* const ostart = (BYTE*)dst; |
3176 | 0 | BYTE* op = ostart; |
3177 | 0 | size_t fhSize = 0; |
3178 | |
|
3179 | 0 | DEBUGLOG(4, "ZSTD_writeEpilogue"); |
3180 | 0 | RETURN_ERROR_IF(cctx->stage == ZSTDcs_created, stage_wrong, "init missing"); |
3181 | | |
3182 | | /* special case : empty frame */ |
3183 | 0 | if (cctx->stage == ZSTDcs_init) { |
3184 | 0 | fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams, 0, 0); |
3185 | 0 | FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed"); |
3186 | 0 | dstCapacity -= fhSize; |
3187 | 0 | op += fhSize; |
3188 | 0 | cctx->stage = ZSTDcs_ongoing; |
3189 | 0 | } |
3190 | | |
3191 | 0 | if (cctx->stage != ZSTDcs_ending) { |
3192 | | /* write one last empty block, make it the "last" block */ |
3193 | 0 | U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw)<<1) + 0; |
3194 | 0 | RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for epilogue"); |
3195 | 0 | MEM_writeLE32(op, cBlockHeader24); |
3196 | 0 | op += ZSTDInternalConstants::ZSTD_blockHeaderSize; |
3197 | 0 | dstCapacity -= ZSTDInternalConstants::ZSTD_blockHeaderSize; |
3198 | 0 | } |
3199 | | |
3200 | 0 | if (cctx->appliedParams.fParams.checksumFlag) { |
3201 | 0 | U32 const checksum = (U32) XXH64_digest(&cctx->xxhState); |
3202 | 0 | RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for checksum"); |
3203 | 0 | DEBUGLOG(4, "ZSTD_writeEpilogue: write checksum : %08X", (unsigned)checksum); |
3204 | 0 | MEM_writeLE32(op, checksum); |
3205 | 0 | op += 4; |
3206 | 0 | } |
3207 | | |
3208 | 0 | cctx->stage = ZSTDcs_created; /* return to "created but no init" status */ |
3209 | 0 | return op-ostart; |
3210 | 0 | } |
3211 | | |
3212 | | size_t ZSTD_compressEnd (ZSTD_CCtx* cctx, |
3213 | | void* dst, size_t dstCapacity, |
3214 | | const void* src, size_t srcSize) |
3215 | 0 | { |
3216 | 0 | size_t endResult; |
3217 | 0 | size_t const cSize = ZSTD_compressContinue_internal(cctx, |
3218 | 0 | dst, dstCapacity, src, srcSize, |
3219 | 0 | 1 /* frame mode */, 1 /* last chunk */); |
3220 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressContinue_internal failed"); |
3221 | 0 | endResult = ZSTD_writeEpilogue(cctx, (char*)dst + cSize, dstCapacity-cSize); |
3222 | 0 | FORWARD_IF_ERROR(endResult, "ZSTD_writeEpilogue failed"); |
3223 | 0 | assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0)); |
3224 | 0 | if (cctx->pledgedSrcSizePlusOne != 0) { /* control src size */ |
3225 | 0 | ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1); |
3226 | 0 | DEBUGLOG(4, "end of frame : controlling src size"); |
3227 | 0 | RETURN_ERROR_IF( |
3228 | 0 | cctx->pledgedSrcSizePlusOne != cctx->consumedSrcSize+1, |
3229 | 0 | srcSize_wrong, |
3230 | 0 | "error : pledgedSrcSize = %u, while realSrcSize = %u", |
3231 | 0 | (unsigned)cctx->pledgedSrcSizePlusOne-1, |
3232 | 0 | (unsigned)cctx->consumedSrcSize); |
3233 | 0 | } |
3234 | 0 | return cSize + endResult; |
3235 | 0 | } |
3236 | | |
3237 | | |
3238 | | static size_t ZSTD_compress_internal (ZSTD_CCtx* cctx, |
3239 | | void* dst, size_t dstCapacity, |
3240 | | const void* src, size_t srcSize, |
3241 | | const void* dict,size_t dictSize, |
3242 | | const ZSTD_parameters* params) |
3243 | 0 | { |
3244 | 0 | ZSTD_CCtx_params const cctxParams = |
3245 | 0 | ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, params); |
3246 | 0 | DEBUGLOG(4, "ZSTD_compress_internal"); |
3247 | 0 | return ZSTD_compress_advanced_internal(cctx, |
3248 | 0 | dst, dstCapacity, |
3249 | 0 | src, srcSize, |
3250 | 0 | dict, dictSize, |
3251 | 0 | &cctxParams); |
3252 | 0 | } |
3253 | | |
3254 | | size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx, |
3255 | | void* dst, size_t dstCapacity, |
3256 | | const void* src, size_t srcSize, |
3257 | | const void* dict,size_t dictSize, |
3258 | | ZSTD_parameters params) |
3259 | 0 | { |
3260 | 0 | DEBUGLOG(4, "ZSTD_compress_advanced"); |
3261 | 0 | FORWARD_IF_ERROR(ZSTD_checkCParams(params.cParams), ""); |
3262 | 0 | return ZSTD_compress_internal(cctx, |
3263 | 0 | dst, dstCapacity, |
3264 | 0 | src, srcSize, |
3265 | 0 | dict, dictSize, |
3266 | 0 | ¶ms); |
3267 | 0 | } |
3268 | | |
3269 | | /* Internal */ |
3270 | | size_t ZSTD_compress_advanced_internal( |
3271 | | ZSTD_CCtx* cctx, |
3272 | | void* dst, size_t dstCapacity, |
3273 | | const void* src, size_t srcSize, |
3274 | | const void* dict,size_t dictSize, |
3275 | | const ZSTD_CCtx_params* params) |
3276 | 0 | { |
3277 | 0 | DEBUGLOG(4, "ZSTD_compress_advanced_internal (srcSize:%u)", (unsigned)srcSize); |
3278 | 0 | FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx, |
3279 | 0 | dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL, |
3280 | 0 | params, srcSize, ZSTDb_not_buffered) , ""); |
3281 | 0 | return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize); |
3282 | 0 | } |
3283 | | |
3284 | | size_t ZSTD_compress_usingDict(ZSTD_CCtx* cctx, |
3285 | | void* dst, size_t dstCapacity, |
3286 | | const void* src, size_t srcSize, |
3287 | | const void* dict, size_t dictSize, |
3288 | | int compressionLevel) |
3289 | 0 | { |
3290 | 0 | ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, srcSize, dict ? dictSize : 0); |
3291 | 0 | ZSTD_CCtx_params cctxParams = ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, ¶ms); |
3292 | 0 | DEBUGLOG(4, "ZSTD_compress_usingDict (srcSize=%u)", (unsigned)srcSize); |
3293 | 0 | assert(params.fParams.contentSizeFlag == 1); |
3294 | 0 | return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, dict, dictSize, &cctxParams); |
3295 | 0 | } |
3296 | | |
3297 | | size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx, |
3298 | | void* dst, size_t dstCapacity, |
3299 | | const void* src, size_t srcSize, |
3300 | | int compressionLevel) |
3301 | 0 | { |
3302 | 0 | DEBUGLOG(4, "ZSTD_compressCCtx (srcSize=%u)", (unsigned)srcSize); |
3303 | 0 | assert(cctx != NULL); |
3304 | 0 | return ZSTD_compress_usingDict(cctx, dst, dstCapacity, src, srcSize, NULL, 0, compressionLevel); |
3305 | 0 | } |
3306 | | |
3307 | | size_t ZSTD_compress(void* dst, size_t dstCapacity, |
3308 | | const void* src, size_t srcSize, |
3309 | | int compressionLevel) |
3310 | 0 | { |
3311 | 0 | size_t result; |
3312 | 0 | ZSTD_CCtx ctxBody; |
3313 | 0 | ZSTD_initCCtx(&ctxBody, ZSTDInternalConstants::ZSTD_defaultCMem); |
3314 | 0 | result = ZSTD_compressCCtx(&ctxBody, dst, dstCapacity, src, srcSize, compressionLevel); |
3315 | 0 | ZSTD_freeCCtxContent(&ctxBody); /* can't free ctxBody itself, as it's on stack; free only heap content */ |
3316 | 0 | return result; |
3317 | 0 | } |
3318 | | |
3319 | | |
3320 | | /* ===== Dictionary API ===== */ |
3321 | | |
3322 | | /*! ZSTD_estimateCDictSize_advanced() : |
3323 | | * Estimate amount of memory that will be needed to create a dictionary with following arguments */ |
3324 | | size_t ZSTD_estimateCDictSize_advanced( |
3325 | | size_t dictSize, ZSTD_compressionParameters cParams, |
3326 | | ZSTD_dictLoadMethod_e dictLoadMethod) |
3327 | 0 | { |
3328 | 0 | DEBUGLOG(5, "sizeof(ZSTD_CDict) : %u", (unsigned)sizeof(ZSTD_CDict)); |
3329 | 0 | return ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) |
3330 | 0 | + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) |
3331 | 0 | + ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0) |
3332 | 0 | + (dictLoadMethod == ZSTD_dlm_byRef ? 0 |
3333 | 0 | : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void *)))); |
3334 | 0 | } |
3335 | | |
3336 | | size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel) |
3337 | 0 | { |
3338 | 0 | ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize); |
3339 | 0 | return ZSTD_estimateCDictSize_advanced(dictSize, cParams, ZSTD_dlm_byCopy); |
3340 | 0 | } |
3341 | | |
3342 | | size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict) |
3343 | 0 | { |
3344 | 0 | if (cdict==NULL) return 0; /* support sizeof on NULL */ |
3345 | 0 | DEBUGLOG(5, "sizeof(*cdict) : %u", (unsigned)sizeof(*cdict)); |
3346 | | /* cdict may be in the workspace */ |
3347 | 0 | return (cdict->workspace.workspace == cdict ? 0 : sizeof(*cdict)) |
3348 | 0 | + ZSTD_cwksp_sizeof(&cdict->workspace); |
3349 | 0 | } |
3350 | | |
3351 | | static size_t ZSTD_initCDict_internal( |
3352 | | ZSTD_CDict* cdict, |
3353 | | const void* dictBuffer, size_t dictSize, |
3354 | | ZSTD_dictLoadMethod_e dictLoadMethod, |
3355 | | ZSTD_dictContentType_e dictContentType, |
3356 | | ZSTD_compressionParameters cParams) |
3357 | 0 | { |
3358 | 0 | DEBUGLOG(3, "ZSTD_initCDict_internal (dictContentType:%u)", (unsigned)dictContentType); |
3359 | 0 | assert(!ZSTD_checkCParams(cParams)); |
3360 | 0 | cdict->matchState.cParams = cParams; |
3361 | 0 | if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dictBuffer) || (!dictSize)) { |
3362 | 0 | cdict->dictContent = dictBuffer; |
3363 | 0 | } else { |
3364 | 0 | void *internalBuffer = ZSTD_cwksp_reserve_object(&cdict->workspace, ZSTD_cwksp_align(dictSize, sizeof(void*))); |
3365 | 0 | RETURN_ERROR_IF(!internalBuffer, memory_allocation, "NULL pointer!"); |
3366 | 0 | cdict->dictContent = internalBuffer; |
3367 | 0 | memcpy(internalBuffer, dictBuffer, dictSize); |
3368 | 0 | } |
3369 | 0 | cdict->dictContentSize = dictSize; |
3370 | |
|
3371 | 0 | cdict->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cdict->workspace, HUF_WORKSPACE_SIZE); |
3372 | | |
3373 | | |
3374 | | /* Reset the state to no dictionary */ |
3375 | 0 | ZSTD_reset_compressedBlockState(&cdict->cBlockState); |
3376 | 0 | FORWARD_IF_ERROR(ZSTD_reset_matchState( |
3377 | 0 | &cdict->matchState, |
3378 | 0 | &cdict->workspace, |
3379 | 0 | &cParams, |
3380 | 0 | ZSTDcrp_makeClean, |
3381 | 0 | ZSTDirp_reset, |
3382 | 0 | ZSTD_resetTarget_CDict), ""); |
3383 | | /* (Maybe) load the dictionary |
3384 | | * Skips loading the dictionary if it is < 8 bytes. |
3385 | | */ |
3386 | 0 | { ZSTD_CCtx_params params; |
3387 | 0 | memset(¶ms, 0, sizeof(params)); |
3388 | 0 | params.compressionLevel = ZSTD_CLEVEL_DEFAULT; |
3389 | 0 | params.fParams.contentSizeFlag = 1; |
3390 | 0 | params.cParams = cParams; |
3391 | 0 | { size_t const dictID = ZSTD_compress_insertDictionary( |
3392 | 0 | &cdict->cBlockState, &cdict->matchState, NULL, &cdict->workspace, |
3393 | 0 | ¶ms, cdict->dictContent, cdict->dictContentSize, |
3394 | 0 | dictContentType, ZSTD_dtlm_full, cdict->entropyWorkspace); |
3395 | 0 | FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed"); |
3396 | 0 | assert(dictID <= (size_t)(U32)-1); |
3397 | 0 | cdict->dictID = (U32)dictID; |
3398 | 0 | } |
3399 | 0 | } |
3400 | | |
3401 | 0 | return 0; |
3402 | 0 | } |
3403 | | |
3404 | | ZSTD_CDict* ZSTD_createCDict_advanced(const void* dictBuffer, size_t dictSize, |
3405 | | ZSTD_dictLoadMethod_e dictLoadMethod, |
3406 | | ZSTD_dictContentType_e dictContentType, |
3407 | | ZSTD_compressionParameters cParams, ZSTD_customMem customMem) |
3408 | 0 | { |
3409 | 0 | DEBUGLOG(3, "ZSTD_createCDict_advanced, mode %u", (unsigned)dictContentType); |
3410 | 0 | if (!customMem.customAlloc ^ !customMem.customFree) return NULL; |
3411 | | |
3412 | 0 | { size_t const workspaceSize = |
3413 | 0 | ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) + |
3414 | 0 | ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) + |
3415 | 0 | ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0) + |
3416 | 0 | (dictLoadMethod == ZSTD_dlm_byRef ? 0 |
3417 | 0 | : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*)))); |
3418 | 0 | void* const workspace = ZSTD_malloc(workspaceSize, customMem); |
3419 | 0 | ZSTD_cwksp ws; |
3420 | 0 | ZSTD_CDict* cdict; |
3421 | |
|
3422 | 0 | if (!workspace) { |
3423 | 0 | ZSTD_free(workspace, customMem); |
3424 | 0 | return NULL; |
3425 | 0 | } |
3426 | | |
3427 | 0 | ZSTD_cwksp_init(&ws, workspace, workspaceSize); |
3428 | |
|
3429 | 0 | cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict)); |
3430 | 0 | assert(cdict != NULL); |
3431 | 0 | ZSTD_cwksp_move(&cdict->workspace, &ws); |
3432 | 0 | cdict->customMem = customMem; |
3433 | 0 | cdict->compressionLevel = 0; /* signals advanced API usage */ |
3434 | |
|
3435 | 0 | if (ZSTD_isError( ZSTD_initCDict_internal(cdict, |
3436 | 0 | dictBuffer, dictSize, |
3437 | 0 | dictLoadMethod, dictContentType, |
3438 | 0 | cParams) )) { |
3439 | 0 | ZSTD_freeCDict(cdict); |
3440 | 0 | return NULL; |
3441 | 0 | } |
3442 | | |
3443 | 0 | return cdict; |
3444 | 0 | } |
3445 | 0 | } |
3446 | | |
3447 | | ZSTD_CDict* ZSTD_createCDict(const void* dict, size_t dictSize, int compressionLevel) |
3448 | 0 | { |
3449 | 0 | ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize); |
3450 | 0 | ZSTD_CDict* cdict = ZSTD_createCDict_advanced(dict, dictSize, |
3451 | 0 | ZSTD_dlm_byCopy, ZSTD_dct_auto, |
3452 | 0 | cParams, ZSTDInternalConstants::ZSTD_defaultCMem); |
3453 | 0 | if (cdict) |
3454 | 0 | cdict->compressionLevel = compressionLevel == 0 ? ZSTD_CLEVEL_DEFAULT : compressionLevel; |
3455 | 0 | return cdict; |
3456 | 0 | } |
3457 | | |
3458 | | ZSTD_CDict* ZSTD_createCDict_byReference(const void* dict, size_t dictSize, int compressionLevel) |
3459 | 0 | { |
3460 | 0 | ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize); |
3461 | 0 | return ZSTD_createCDict_advanced(dict, dictSize, |
3462 | 0 | ZSTD_dlm_byRef, ZSTD_dct_auto, |
3463 | 0 | cParams, ZSTDInternalConstants::ZSTD_defaultCMem); |
3464 | 0 | } |
3465 | | |
3466 | | size_t ZSTD_freeCDict(ZSTD_CDict* cdict) |
3467 | 0 | { |
3468 | 0 | if (cdict==NULL) return 0; /* support free on NULL */ |
3469 | 0 | { ZSTD_customMem const cMem = cdict->customMem; |
3470 | 0 | int cdictInWorkspace = ZSTD_cwksp_owns_buffer(&cdict->workspace, cdict); |
3471 | 0 | ZSTD_cwksp_free(&cdict->workspace, cMem); |
3472 | 0 | if (!cdictInWorkspace) { |
3473 | 0 | ZSTD_free(cdict, cMem); |
3474 | 0 | } |
3475 | 0 | return 0; |
3476 | 0 | } |
3477 | 0 | } |
3478 | | |
3479 | | /*! ZSTD_initStaticCDict_advanced() : |
3480 | | * Generate a digested dictionary in provided memory area. |
3481 | | * workspace: The memory area to emplace the dictionary into. |
3482 | | * Provided pointer must 8-bytes aligned. |
3483 | | * It must outlive dictionary usage. |
3484 | | * workspaceSize: Use ZSTD_estimateCDictSize() |
3485 | | * to determine how large workspace must be. |
3486 | | * cParams : use ZSTD_getCParams() to transform a compression level |
3487 | | * into its relevants cParams. |
3488 | | * @return : pointer to ZSTD_CDict*, or NULL if error (size too small) |
3489 | | * Note : there is no corresponding "free" function. |
3490 | | * Since workspace was allocated externally, it must be freed externally. |
3491 | | */ |
3492 | | const ZSTD_CDict* ZSTD_initStaticCDict( |
3493 | | void* workspace, size_t workspaceSize, |
3494 | | const void* dict, size_t dictSize, |
3495 | | ZSTD_dictLoadMethod_e dictLoadMethod, |
3496 | | ZSTD_dictContentType_e dictContentType, |
3497 | | ZSTD_compressionParameters cParams) |
3498 | 0 | { |
3499 | 0 | size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0); |
3500 | 0 | size_t const neededSize = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) |
3501 | 0 | + (dictLoadMethod == ZSTD_dlm_byRef ? 0 |
3502 | 0 | : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*)))) |
3503 | 0 | + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) |
3504 | 0 | + matchStateSize; |
3505 | 0 | ZSTD_CDict* cdict; |
3506 | |
|
3507 | 0 | if ((size_t)workspace & 7) return NULL; /* 8-aligned */ |
3508 | | |
3509 | 0 | { |
3510 | 0 | ZSTD_cwksp ws; |
3511 | 0 | ZSTD_cwksp_init(&ws, workspace, workspaceSize); |
3512 | 0 | cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict)); |
3513 | 0 | if (cdict == NULL) return NULL; |
3514 | 0 | ZSTD_cwksp_move(&cdict->workspace, &ws); |
3515 | 0 | } |
3516 | | |
3517 | 0 | DEBUGLOG(4, "(workspaceSize < neededSize) : (%u < %u) => %u", |
3518 | 0 | (unsigned)workspaceSize, (unsigned)neededSize, (unsigned)(workspaceSize < neededSize)); |
3519 | 0 | if (workspaceSize < neededSize) return NULL; |
3520 | | |
3521 | 0 | if (ZSTD_isError( ZSTD_initCDict_internal(cdict, |
3522 | 0 | dict, dictSize, |
3523 | 0 | dictLoadMethod, dictContentType, |
3524 | 0 | cParams) )) |
3525 | 0 | return NULL; |
3526 | | |
3527 | 0 | return cdict; |
3528 | 0 | } |
3529 | | |
3530 | | ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict) |
3531 | 0 | { |
3532 | 0 | assert(cdict != NULL); |
3533 | 0 | return cdict->matchState.cParams; |
3534 | 0 | } |
3535 | | |
3536 | | /* ZSTD_compressBegin_usingCDict_advanced() : |
3537 | | * cdict must be != NULL */ |
3538 | | size_t ZSTD_compressBegin_usingCDict_advanced( |
3539 | | ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict, |
3540 | | ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize) |
3541 | 0 | { |
3542 | 0 | DEBUGLOG(4, "ZSTD_compressBegin_usingCDict_advanced"); |
3543 | 0 | RETURN_ERROR_IF(cdict==NULL, dictionary_wrong, "NULL pointer!"); |
3544 | 0 | { ZSTD_CCtx_params params = cctx->requestedParams; |
3545 | 0 | params.cParams = ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF |
3546 | 0 | || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER |
3547 | 0 | || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN |
3548 | 0 | || cdict->compressionLevel == 0 ) |
3549 | 0 | && (params.attachDictPref != ZSTD_dictForceLoad) ? |
3550 | 0 | ZSTD_getCParamsFromCDict(cdict) |
3551 | 0 | : ZSTD_getCParams(cdict->compressionLevel, |
3552 | 0 | pledgedSrcSize, |
3553 | 0 | cdict->dictContentSize); |
3554 | | /* Increase window log to fit the entire dictionary and source if the |
3555 | | * source size is known. Limit the increase to 19, which is the |
3556 | | * window log for compression level 1 with the largest source size. |
3557 | | */ |
3558 | 0 | if (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN) { |
3559 | 0 | U32 const limitedSrcSize = (U32)MIN(pledgedSrcSize, 1U << 19); |
3560 | 0 | U32 const limitedSrcLog = limitedSrcSize > 1 ? ZSTD_highbit32(limitedSrcSize - 1) + 1 : 1; |
3561 | 0 | params.cParams.windowLog = MAX(params.cParams.windowLog, limitedSrcLog); |
3562 | 0 | } |
3563 | 0 | params.fParams = fParams; |
3564 | 0 | return ZSTD_compressBegin_internal(cctx, |
3565 | 0 | NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast, |
3566 | 0 | cdict, |
3567 | 0 | ¶ms, pledgedSrcSize, |
3568 | 0 | ZSTDb_not_buffered); |
3569 | 0 | } |
3570 | 0 | } |
3571 | | |
3572 | | /* ZSTD_compressBegin_usingCDict() : |
3573 | | * pledgedSrcSize=0 means "unknown" |
3574 | | * if pledgedSrcSize>0, it will enable contentSizeFlag */ |
3575 | | size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict) |
3576 | 0 | { |
3577 | 0 | ZSTD_frameParameters const fParams = { 0 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ }; |
3578 | 0 | DEBUGLOG(4, "ZSTD_compressBegin_usingCDict : dictIDFlag == %u", !fParams.noDictIDFlag); |
3579 | 0 | return ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, ZSTD_CONTENTSIZE_UNKNOWN); |
3580 | 0 | } |
3581 | | |
3582 | | size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx, |
3583 | | void* dst, size_t dstCapacity, |
3584 | | const void* src, size_t srcSize, |
3585 | | const ZSTD_CDict* cdict, ZSTD_frameParameters fParams) |
3586 | 0 | { |
3587 | 0 | FORWARD_IF_ERROR(ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, srcSize), ""); /* will check if cdict != NULL */ |
3588 | 0 | return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize); |
3589 | 0 | } |
3590 | | |
3591 | | /*! ZSTD_compress_usingCDict() : |
3592 | | * Compression using a digested Dictionary. |
3593 | | * Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times. |
3594 | | * Note that compression parameters are decided at CDict creation time |
3595 | | * while frame parameters are hardcoded */ |
3596 | | size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx, |
3597 | | void* dst, size_t dstCapacity, |
3598 | | const void* src, size_t srcSize, |
3599 | | const ZSTD_CDict* cdict) |
3600 | 0 | { |
3601 | 0 | ZSTD_frameParameters const fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ }; |
3602 | 0 | return ZSTD_compress_usingCDict_advanced(cctx, dst, dstCapacity, src, srcSize, cdict, fParams); |
3603 | 0 | } |
3604 | | |
3605 | | |
3606 | | |
3607 | | /* ****************************************************************** |
3608 | | * Streaming |
3609 | | ********************************************************************/ |
3610 | | |
3611 | | ZSTD_CStream* ZSTD_createCStream(void) |
3612 | 0 | { |
3613 | 0 | DEBUGLOG(3, "ZSTD_createCStream"); |
3614 | 0 | return ZSTD_createCStream_advanced(ZSTDInternalConstants::ZSTD_defaultCMem); |
3615 | 0 | } |
3616 | | |
3617 | | ZSTD_CStream* ZSTD_initStaticCStream(void *workspace, size_t workspaceSize) |
3618 | 0 | { |
3619 | 0 | return ZSTD_initStaticCCtx(workspace, workspaceSize); |
3620 | 0 | } |
3621 | | |
3622 | | ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem) |
3623 | 0 | { /* CStream and CCtx are now same object */ |
3624 | 0 | return ZSTD_createCCtx_advanced(customMem); |
3625 | 0 | } |
3626 | | |
3627 | | size_t ZSTD_freeCStream(ZSTD_CStream* zcs) |
3628 | 0 | { |
3629 | 0 | return ZSTD_freeCCtx(zcs); /* same object */ |
3630 | 0 | } |
3631 | | |
3632 | | |
3633 | | |
3634 | | /*====== Initialization ======*/ |
3635 | | |
3636 | 0 | size_t ZSTD_CStreamInSize(void) { return ZSTD_BLOCKSIZE_MAX; } |
3637 | | |
3638 | | size_t ZSTD_CStreamOutSize(void) |
3639 | 0 | { |
3640 | 0 | return ZSTD_compressBound(ZSTD_BLOCKSIZE_MAX) + ZSTDInternalConstants::ZSTD_blockHeaderSize + 4 /* 32-bits hash */ ; |
3641 | 0 | } |
3642 | | |
3643 | | static size_t ZSTD_resetCStream_internal(ZSTD_CStream* cctx, |
3644 | | const void* const dict, size_t const dictSize, ZSTD_dictContentType_e const dictContentType, |
3645 | | const ZSTD_CDict* const cdict, |
3646 | | ZSTD_CCtx_params params, unsigned long long const pledgedSrcSize) |
3647 | 0 | { |
3648 | 0 | DEBUGLOG(4, "ZSTD_resetCStream_internal"); |
3649 | | /* Finalize the compression parameters */ |
3650 | 0 | params.cParams = ZSTD_getCParamsFromCCtxParams(¶ms, pledgedSrcSize, dictSize); |
3651 | | /* params are supposed to be fully validated at this point */ |
3652 | 0 | assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams))); |
3653 | 0 | assert(!((dict) && (cdict))); /* either dict or cdict, not both */ |
3654 | |
|
3655 | 0 | FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx, |
3656 | 0 | dict, dictSize, dictContentType, ZSTD_dtlm_fast, |
3657 | 0 | cdict, |
3658 | 0 | ¶ms, pledgedSrcSize, |
3659 | 0 | ZSTDb_buffered) , ""); |
3660 | |
|
3661 | 0 | cctx->inToCompress = 0; |
3662 | 0 | cctx->inBuffPos = 0; |
3663 | 0 | cctx->inBuffTarget = cctx->blockSize |
3664 | 0 | + (cctx->blockSize == pledgedSrcSize); /* for small input: avoid automatic flush on reaching end of block, since it would require to add a 3-bytes null block to end frame */ |
3665 | 0 | cctx->outBuffContentSize = cctx->outBuffFlushedSize = 0; |
3666 | 0 | cctx->streamStage = zcss_load; |
3667 | 0 | cctx->frameEnded = 0; |
3668 | 0 | return 0; /* ready to go */ |
3669 | 0 | } |
3670 | | |
3671 | | /* ZSTD_resetCStream(): |
3672 | | * pledgedSrcSize == 0 means "unknown" */ |
3673 | | size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pss) |
3674 | 0 | { |
3675 | | /* temporary : 0 interpreted as "unknown" during transition period. |
3676 | | * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN. |
3677 | | * 0 will be interpreted as "empty" in the future. |
3678 | | */ |
3679 | 0 | U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss; |
3680 | 0 | DEBUGLOG(4, "ZSTD_resetCStream: pledgedSrcSize = %u", (unsigned)pledgedSrcSize); |
3681 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3682 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , ""); |
3683 | 0 | return 0; |
3684 | 0 | } |
3685 | | |
3686 | | /*! ZSTD_initCStream_internal() : |
3687 | | * Note : for lib/compress only. Used by zstdmt_compress.c. |
3688 | | * Assumption 1 : params are valid |
3689 | | * Assumption 2 : either dict, or cdict, is defined, not both */ |
3690 | | size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs, |
3691 | | const void* dict, size_t dictSize, const ZSTD_CDict* cdict, |
3692 | | const ZSTD_CCtx_params* params, |
3693 | | unsigned long long pledgedSrcSize) |
3694 | 0 | { |
3695 | 0 | DEBUGLOG(4, "ZSTD_initCStream_internal"); |
3696 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3697 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , ""); |
3698 | 0 | assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams))); |
3699 | 0 | zcs->requestedParams = *params; |
3700 | 0 | assert(!((dict) && (cdict))); /* either dict or cdict, not both */ |
3701 | 0 | if (dict) { |
3702 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , ""); |
3703 | 0 | } else { |
3704 | | /* Dictionary is cleared if !cdict */ |
3705 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , ""); |
3706 | 0 | } |
3707 | 0 | return 0; |
3708 | 0 | } |
3709 | | |
3710 | | /* ZSTD_initCStream_usingCDict_advanced() : |
3711 | | * same as ZSTD_initCStream_usingCDict(), with control over frame parameters */ |
3712 | | size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs, |
3713 | | const ZSTD_CDict* cdict, |
3714 | | ZSTD_frameParameters fParams, |
3715 | | unsigned long long pledgedSrcSize) |
3716 | 0 | { |
3717 | 0 | DEBUGLOG(4, "ZSTD_initCStream_usingCDict_advanced"); |
3718 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3719 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , ""); |
3720 | 0 | zcs->requestedParams.fParams = fParams; |
3721 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , ""); |
3722 | 0 | return 0; |
3723 | 0 | } |
3724 | | |
3725 | | /* note : cdict must outlive compression session */ |
3726 | | size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict) |
3727 | 0 | { |
3728 | 0 | DEBUGLOG(4, "ZSTD_initCStream_usingCDict"); |
3729 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3730 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , ""); |
3731 | 0 | return 0; |
3732 | 0 | } |
3733 | | |
3734 | | |
3735 | | /* ZSTD_initCStream_advanced() : |
3736 | | * pledgedSrcSize must be exact. |
3737 | | * if srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN. |
3738 | | * dict is loaded with default parameters ZSTD_dct_auto and ZSTD_dlm_byCopy. */ |
3739 | | size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs, |
3740 | | const void* dict, size_t dictSize, |
3741 | | ZSTD_parameters params, unsigned long long pss) |
3742 | 0 | { |
3743 | | /* for compatibility with older programs relying on this behavior. |
3744 | | * Users should now specify ZSTD_CONTENTSIZE_UNKNOWN. |
3745 | | * This line will be removed in the future. |
3746 | | */ |
3747 | 0 | U64 const pledgedSrcSize = (pss==0 && params.fParams.contentSizeFlag==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss; |
3748 | 0 | DEBUGLOG(4, "ZSTD_initCStream_advanced"); |
3749 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3750 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , ""); |
3751 | 0 | FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , ""); |
3752 | 0 | zcs->requestedParams = ZSTD_assignParamsToCCtxParams(&zcs->requestedParams, ¶ms); |
3753 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , ""); |
3754 | 0 | return 0; |
3755 | 0 | } |
3756 | | |
3757 | | size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs, const void* dict, size_t dictSize, int compressionLevel) |
3758 | 0 | { |
3759 | 0 | DEBUGLOG(4, "ZSTD_initCStream_usingDict"); |
3760 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3761 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , ""); |
3762 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , ""); |
3763 | 0 | return 0; |
3764 | 0 | } |
3765 | | |
3766 | | size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs, int compressionLevel, unsigned long long pss) |
3767 | 0 | { |
3768 | | /* temporary : 0 interpreted as "unknown" during transition period. |
3769 | | * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN. |
3770 | | * 0 will be interpreted as "empty" in the future. |
3771 | | */ |
3772 | 0 | U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss; |
3773 | 0 | DEBUGLOG(4, "ZSTD_initCStream_srcSize"); |
3774 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3775 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , ""); |
3776 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , ""); |
3777 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , ""); |
3778 | 0 | return 0; |
3779 | 0 | } |
3780 | | |
3781 | | size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel) |
3782 | 0 | { |
3783 | 0 | DEBUGLOG(4, "ZSTD_initCStream"); |
3784 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , ""); |
3785 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , ""); |
3786 | 0 | FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , ""); |
3787 | 0 | return 0; |
3788 | 0 | } |
3789 | | |
3790 | | /*====== Compression ======*/ |
3791 | | |
3792 | | static size_t ZSTD_nextInputSizeHint(const ZSTD_CCtx* cctx) |
3793 | 0 | { |
3794 | 0 | size_t hintInSize = cctx->inBuffTarget - cctx->inBuffPos; |
3795 | 0 | if (hintInSize==0) hintInSize = cctx->blockSize; |
3796 | 0 | return hintInSize; |
3797 | 0 | } |
3798 | | |
3799 | | /** ZSTD_compressStream_generic(): |
3800 | | * internal function for all *compressStream*() variants |
3801 | | * non-static, because can be called from zstdmt_compress.c |
3802 | | * @return : hint size for next input */ |
3803 | | static size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs, |
3804 | | ZSTD_outBuffer* output, |
3805 | | ZSTD_inBuffer* input, |
3806 | | ZSTD_EndDirective const flushMode) |
3807 | 0 | { |
3808 | 0 | const char* const istart = (const char*)input->src; |
3809 | 0 | const char* const iend = input->size != 0 ? istart + input->size : istart; |
3810 | 0 | const char* ip = input->pos != 0 ? istart + input->pos : istart; |
3811 | 0 | char* const ostart = (char*)output->dst; |
3812 | 0 | char* const oend = output->size != 0 ? ostart + output->size : ostart; |
3813 | 0 | char* op = output->pos != 0 ? ostart + output->pos : ostart; |
3814 | 0 | U32 someMoreWork = 1; |
3815 | | |
3816 | | /* check expectations */ |
3817 | 0 | DEBUGLOG(5, "ZSTD_compressStream_generic, flush=%u", (unsigned)flushMode); |
3818 | 0 | assert(zcs->inBuff != NULL); |
3819 | 0 | assert(zcs->inBuffSize > 0); |
3820 | 0 | assert(zcs->outBuff != NULL); |
3821 | 0 | assert(zcs->outBuffSize > 0); |
3822 | 0 | assert(output->pos <= output->size); |
3823 | 0 | assert(input->pos <= input->size); |
3824 | |
|
3825 | 0 | while (someMoreWork) { |
3826 | 0 | switch(zcs->streamStage) |
3827 | 0 | { |
3828 | 0 | case zcss_init: |
3829 | 0 | RETURN_ERROR(init_missing, "call ZSTD_initCStream() first!"); |
3830 | |
|
3831 | 0 | case zcss_load: |
3832 | 0 | if ( (flushMode == ZSTD_e_end) |
3833 | 0 | && ((size_t)(oend-op) >= ZSTD_compressBound(iend-ip)) /* enough dstCapacity */ |
3834 | 0 | && (zcs->inBuffPos == 0) ) { |
3835 | | /* shortcut to compression pass directly into output buffer */ |
3836 | 0 | size_t const cSize = ZSTD_compressEnd(zcs, |
3837 | 0 | op, oend-op, ip, iend-ip); |
3838 | 0 | DEBUGLOG(4, "ZSTD_compressEnd : cSize=%u", (unsigned)cSize); |
3839 | 0 | FORWARD_IF_ERROR(cSize, "ZSTD_compressEnd failed"); |
3840 | 0 | ip = iend; |
3841 | 0 | op += cSize; |
3842 | 0 | zcs->frameEnded = 1; |
3843 | 0 | ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only); |
3844 | 0 | someMoreWork = 0; break; |
3845 | 0 | } |
3846 | | /* complete loading into inBuffer */ |
3847 | 0 | { size_t const toLoad = zcs->inBuffTarget - zcs->inBuffPos; |
3848 | 0 | size_t const loaded = ZSTD_limitCopy( |
3849 | 0 | zcs->inBuff + zcs->inBuffPos, toLoad, |
3850 | 0 | ip, iend-ip); |
3851 | 0 | zcs->inBuffPos += loaded; |
3852 | 0 | if (loaded != 0) |
3853 | 0 | ip += loaded; |
3854 | 0 | if ( (flushMode == ZSTD_e_continue) |
3855 | 0 | && (zcs->inBuffPos < zcs->inBuffTarget) ) { |
3856 | | /* not enough input to fill full block : stop here */ |
3857 | 0 | someMoreWork = 0; break; |
3858 | 0 | } |
3859 | 0 | if ( (flushMode == ZSTD_e_flush) |
3860 | 0 | && (zcs->inBuffPos == zcs->inToCompress) ) { |
3861 | | /* empty */ |
3862 | 0 | someMoreWork = 0; break; |
3863 | 0 | } |
3864 | 0 | } |
3865 | | /* compress current block (note : this stage cannot be stopped in the middle) */ |
3866 | 0 | DEBUGLOG(5, "stream compression stage (flushMode==%u)", flushMode); |
3867 | 0 | { void* cDst; |
3868 | 0 | size_t cSize; |
3869 | 0 | size_t const iSize = zcs->inBuffPos - zcs->inToCompress; |
3870 | 0 | size_t oSize = oend-op; |
3871 | 0 | unsigned const lastBlock = (flushMode == ZSTD_e_end) && (ip==iend); |
3872 | 0 | if (oSize >= ZSTD_compressBound(iSize)) |
3873 | 0 | cDst = op; /* compress into output buffer, to skip flush stage */ |
3874 | 0 | else |
3875 | 0 | cDst = zcs->outBuff, oSize = zcs->outBuffSize; |
3876 | 0 | cSize = lastBlock ? |
3877 | 0 | ZSTD_compressEnd(zcs, cDst, oSize, |
3878 | 0 | zcs->inBuff + zcs->inToCompress, iSize) : |
3879 | 0 | ZSTD_compressContinue(zcs, cDst, oSize, |
3880 | 0 | zcs->inBuff + zcs->inToCompress, iSize); |
3881 | 0 | FORWARD_IF_ERROR(cSize, "%s", lastBlock ? "ZSTD_compressEnd failed" : "ZSTD_compressContinue failed"); |
3882 | 0 | zcs->frameEnded = lastBlock; |
3883 | | /* prepare next block */ |
3884 | 0 | zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize; |
3885 | 0 | if (zcs->inBuffTarget > zcs->inBuffSize) |
3886 | 0 | zcs->inBuffPos = 0, zcs->inBuffTarget = zcs->blockSize; |
3887 | 0 | DEBUGLOG(5, "inBuffTarget:%u / inBuffSize:%u", |
3888 | 0 | (unsigned)zcs->inBuffTarget, (unsigned)zcs->inBuffSize); |
3889 | 0 | if (!lastBlock) |
3890 | 0 | assert(zcs->inBuffTarget <= zcs->inBuffSize); |
3891 | 0 | zcs->inToCompress = zcs->inBuffPos; |
3892 | 0 | if (cDst == op) { /* no need to flush */ |
3893 | 0 | op += cSize; |
3894 | 0 | if (zcs->frameEnded) { |
3895 | 0 | DEBUGLOG(5, "Frame completed directly in outBuffer"); |
3896 | 0 | someMoreWork = 0; |
3897 | 0 | ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only); |
3898 | 0 | } |
3899 | 0 | break; |
3900 | 0 | } |
3901 | 0 | zcs->outBuffContentSize = cSize; |
3902 | 0 | zcs->outBuffFlushedSize = 0; |
3903 | 0 | zcs->streamStage = zcss_flush; /* pass-through to flush stage */ |
3904 | 0 | } |
3905 | | /* fall-through */ |
3906 | 0 | case zcss_flush: |
3907 | 0 | DEBUGLOG(5, "flush stage"); |
3908 | 0 | { size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize; |
3909 | 0 | size_t const flushed = ZSTD_limitCopy(op, (size_t)(oend-op), |
3910 | 0 | zcs->outBuff + zcs->outBuffFlushedSize, toFlush); |
3911 | 0 | DEBUGLOG(5, "toFlush: %u into %u ==> flushed: %u", |
3912 | 0 | (unsigned)toFlush, (unsigned)(oend-op), (unsigned)flushed); |
3913 | 0 | if (flushed) |
3914 | 0 | op += flushed; |
3915 | 0 | zcs->outBuffFlushedSize += flushed; |
3916 | 0 | if (toFlush!=flushed) { |
3917 | | /* flush not fully completed, presumably because dst is too small */ |
3918 | 0 | assert(op==oend); |
3919 | 0 | someMoreWork = 0; |
3920 | 0 | break; |
3921 | 0 | } |
3922 | 0 | zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0; |
3923 | 0 | if (zcs->frameEnded) { |
3924 | 0 | DEBUGLOG(5, "Frame completed on flush"); |
3925 | 0 | someMoreWork = 0; |
3926 | 0 | ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only); |
3927 | 0 | break; |
3928 | 0 | } |
3929 | 0 | zcs->streamStage = zcss_load; |
3930 | 0 | break; |
3931 | 0 | } |
3932 | | |
3933 | 0 | default: /* impossible */ |
3934 | 0 | assert(0); |
3935 | 0 | } |
3936 | 0 | } |
3937 | | |
3938 | 0 | input->pos = ip - istart; |
3939 | 0 | output->pos = op - ostart; |
3940 | 0 | if (zcs->frameEnded) return 0; |
3941 | 0 | return ZSTD_nextInputSizeHint(zcs); |
3942 | 0 | } |
3943 | | |
3944 | | static size_t ZSTD_nextInputSizeHint_MTorST(const ZSTD_CCtx* cctx) |
3945 | 0 | { |
3946 | | #ifdef ZSTD_MULTITHREAD |
3947 | | if (cctx->appliedParams.nbWorkers >= 1) { |
3948 | | assert(cctx->mtctx != NULL); |
3949 | | return ZSTDMT_nextInputSizeHint(cctx->mtctx); |
3950 | | } |
3951 | | #endif |
3952 | 0 | return ZSTD_nextInputSizeHint(cctx); |
3953 | |
|
3954 | 0 | } |
3955 | | |
3956 | | size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input) |
3957 | 0 | { |
3958 | 0 | FORWARD_IF_ERROR( ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue) , ""); |
3959 | 0 | return ZSTD_nextInputSizeHint_MTorST(zcs); |
3960 | 0 | } |
3961 | | |
3962 | | |
3963 | | size_t ZSTD_compressStream2( ZSTD_CCtx* cctx, |
3964 | | ZSTD_outBuffer* output, |
3965 | | ZSTD_inBuffer* input, |
3966 | | ZSTD_EndDirective endOp) |
3967 | 0 | { |
3968 | 0 | DEBUGLOG(5, "ZSTD_compressStream2, endOp=%u ", (unsigned)endOp); |
3969 | | /* check conditions */ |
3970 | 0 | RETURN_ERROR_IF(output->pos > output->size, GENERIC, "invalid buffer"); |
3971 | 0 | RETURN_ERROR_IF(input->pos > input->size, GENERIC, "invalid buffer"); |
3972 | 0 | assert(cctx!=NULL); |
3973 | | |
3974 | | /* transparent initialization stage */ |
3975 | 0 | if (cctx->streamStage == zcss_init) { |
3976 | 0 | ZSTD_CCtx_params params = cctx->requestedParams; |
3977 | 0 | ZSTD_prefixDict const prefixDict = cctx->prefixDict; |
3978 | 0 | FORWARD_IF_ERROR( ZSTD_initLocalDict(cctx) , ""); /* Init the local dict if present. */ |
3979 | 0 | memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict)); /* single usage */ |
3980 | 0 | assert(prefixDict.dict==NULL || cctx->cdict==NULL); /* only one can be set */ |
3981 | 0 | DEBUGLOG(4, "ZSTD_compressStream2 : transparent init stage"); |
3982 | 0 | if (endOp == ZSTD_e_end) cctx->pledgedSrcSizePlusOne = input->size + 1; /* auto-fix pledgedSrcSize */ |
3983 | 0 | params.cParams = ZSTD_getCParamsFromCCtxParams( |
3984 | 0 | &cctx->requestedParams, cctx->pledgedSrcSizePlusOne-1, 0 /*dictSize*/); |
3985 | | |
3986 | |
|
3987 | | #ifdef ZSTD_MULTITHREAD |
3988 | | if ((cctx->pledgedSrcSizePlusOne-1) <= ZSTDMT_JOBSIZE_MIN) { |
3989 | | params.nbWorkers = 0; /* do not invoke multi-threading when src size is too small */ |
3990 | | } |
3991 | | if (params.nbWorkers > 0) { |
3992 | | /* mt context creation */ |
3993 | | if (cctx->mtctx == NULL) { |
3994 | | DEBUGLOG(4, "ZSTD_compressStream2: creating new mtctx for nbWorkers=%u", |
3995 | | params.nbWorkers); |
3996 | | cctx->mtctx = ZSTDMT_createCCtx_advanced((U32)params.nbWorkers, cctx->customMem); |
3997 | | RETURN_ERROR_IF(cctx->mtctx == NULL, memory_allocation, "NULL pointer!"); |
3998 | | } |
3999 | | /* mt compression */ |
4000 | | DEBUGLOG(4, "call ZSTDMT_initCStream_internal as nbWorkers=%u", params.nbWorkers); |
4001 | | FORWARD_IF_ERROR( ZSTDMT_initCStream_internal( |
4002 | | cctx->mtctx, |
4003 | | prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType, |
4004 | | cctx->cdict, params, cctx->pledgedSrcSizePlusOne-1) , ""); |
4005 | | cctx->streamStage = zcss_load; |
4006 | | cctx->appliedParams.nbWorkers = params.nbWorkers; |
4007 | | } else |
4008 | | #endif |
4009 | 0 | { FORWARD_IF_ERROR( ZSTD_resetCStream_internal(cctx, |
4010 | 0 | prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType, |
4011 | 0 | cctx->cdict, |
4012 | 0 | params, cctx->pledgedSrcSizePlusOne-1) , ""); |
4013 | 0 | assert(cctx->streamStage == zcss_load); |
4014 | 0 | assert(cctx->appliedParams.nbWorkers == 0); |
4015 | 0 | } } |
4016 | | /* end of transparent initialization stage */ |
4017 | | |
4018 | | /* compression stage */ |
4019 | | #ifdef ZSTD_MULTITHREAD |
4020 | | if (cctx->appliedParams.nbWorkers > 0) { |
4021 | | int const forceMaxProgress = (endOp == ZSTD_e_flush || endOp == ZSTD_e_end); |
4022 | | size_t flushMin; |
4023 | | assert(forceMaxProgress || endOp == ZSTD_e_continue /* Protection for a new flush type */); |
4024 | | if (cctx->cParamsChanged) { |
4025 | | ZSTDMT_updateCParams_whileCompressing(cctx->mtctx, &cctx->requestedParams); |
4026 | | cctx->cParamsChanged = 0; |
4027 | | } |
4028 | | do { |
4029 | | flushMin = ZSTDMT_compressStream_generic(cctx->mtctx, output, input, endOp); |
4030 | | if ( ZSTD_isError(flushMin) |
4031 | | || (endOp == ZSTD_e_end && flushMin == 0) ) { /* compression completed */ |
4032 | | ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only); |
4033 | | } |
4034 | | FORWARD_IF_ERROR(flushMin, "ZSTDMT_compressStream_generic failed"); |
4035 | | } while (forceMaxProgress && flushMin != 0 && output->pos < output->size); |
4036 | | DEBUGLOG(5, "completed ZSTD_compressStream2 delegating to ZSTDMT_compressStream_generic"); |
4037 | | /* Either we don't require maximum forward progress, we've finished the |
4038 | | * flush, or we are out of output space. |
4039 | | */ |
4040 | | assert(!forceMaxProgress || flushMin == 0 || output->pos == output->size); |
4041 | | return flushMin; |
4042 | | } |
4043 | | #endif |
4044 | 0 | FORWARD_IF_ERROR( ZSTD_compressStream_generic(cctx, output, input, endOp) , ""); |
4045 | 0 | DEBUGLOG(5, "completed ZSTD_compressStream2"); |
4046 | 0 | return cctx->outBuffContentSize - cctx->outBuffFlushedSize; /* remaining to flush */ |
4047 | 0 | } |
4048 | | |
4049 | | size_t ZSTD_compressStream2_simpleArgs ( |
4050 | | ZSTD_CCtx* cctx, |
4051 | | void* dst, size_t dstCapacity, size_t* dstPos, |
4052 | | const void* src, size_t srcSize, size_t* srcPos, |
4053 | | ZSTD_EndDirective endOp) |
4054 | 0 | { |
4055 | 0 | ZSTD_outBuffer output = { dst, dstCapacity, *dstPos }; |
4056 | 0 | ZSTD_inBuffer input = { src, srcSize, *srcPos }; |
4057 | | /* ZSTD_compressStream2() will check validity of dstPos and srcPos */ |
4058 | 0 | size_t const cErr = ZSTD_compressStream2(cctx, &output, &input, endOp); |
4059 | 0 | *dstPos = output.pos; |
4060 | 0 | *srcPos = input.pos; |
4061 | 0 | return cErr; |
4062 | 0 | } |
4063 | | |
4064 | | size_t ZSTD_compress2(ZSTD_CCtx* cctx, |
4065 | | void* dst, size_t dstCapacity, |
4066 | | const void* src, size_t srcSize) |
4067 | 0 | { |
4068 | 0 | DEBUGLOG(4, "ZSTD_compress2 (srcSize=%u)", (unsigned)srcSize); |
4069 | 0 | ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only); |
4070 | 0 | { size_t oPos = 0; |
4071 | 0 | size_t iPos = 0; |
4072 | 0 | size_t const result = ZSTD_compressStream2_simpleArgs(cctx, |
4073 | 0 | dst, dstCapacity, &oPos, |
4074 | 0 | src, srcSize, &iPos, |
4075 | 0 | ZSTD_e_end); |
4076 | 0 | FORWARD_IF_ERROR(result, "ZSTD_compressStream2_simpleArgs failed"); |
4077 | 0 | if (result != 0) { /* compression not completed, due to lack of output space */ |
4078 | 0 | assert(oPos == dstCapacity); |
4079 | 0 | RETURN_ERROR(dstSize_tooSmall, ""); |
4080 | 0 | } |
4081 | 0 | assert(iPos == srcSize); /* all input is expected consumed */ |
4082 | 0 | return oPos; |
4083 | 0 | } |
4084 | 0 | } |
4085 | | |
4086 | | /*====== Finalize ======*/ |
4087 | | |
4088 | | /*! ZSTD_flushStream() : |
4089 | | * @return : amount of data remaining to flush */ |
4090 | | size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output) |
4091 | 0 | { |
4092 | 0 | ZSTD_inBuffer input = { NULL, 0, 0 }; |
4093 | 0 | return ZSTD_compressStream2(zcs, output, &input, ZSTD_e_flush); |
4094 | 0 | } |
4095 | | |
4096 | | |
4097 | | size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output) |
4098 | 0 | { |
4099 | 0 | ZSTD_inBuffer input = { NULL, 0, 0 }; |
4100 | 0 | size_t const remainingToFlush = ZSTD_compressStream2(zcs, output, &input, ZSTD_e_end); |
4101 | 0 | FORWARD_IF_ERROR( remainingToFlush , "ZSTD_compressStream2 failed"); |
4102 | 0 | if (zcs->appliedParams.nbWorkers > 0) return remainingToFlush; /* minimal estimation */ |
4103 | | /* single thread mode : attempt to calculate remaining to flush more precisely */ |
4104 | 0 | { size_t const lastBlockSize = zcs->frameEnded ? 0 : ZSTD_BLOCKHEADERSIZE; |
4105 | 0 | size_t const checksumSize = (size_t)(zcs->frameEnded ? 0 : zcs->appliedParams.fParams.checksumFlag * 4); |
4106 | 0 | size_t const toFlush = remainingToFlush + lastBlockSize + checksumSize; |
4107 | 0 | DEBUGLOG(4, "ZSTD_endStream : remaining to flush : %u", (unsigned)toFlush); |
4108 | 0 | return toFlush; |
4109 | 0 | } |
4110 | 0 | } |
4111 | | |
4112 | | |
4113 | | /*-===== Pre-defined compression levels =====-*/ |
4114 | | |
4115 | 0 | #define ZSTD_MAX_CLEVEL 22 |
4116 | 0 | int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; } |
4117 | 0 | int ZSTD_minCLevel(void) { return (int)-ZSTD_TARGETLENGTH_MAX; } |
4118 | | |
4119 | | static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL+1] = { |
4120 | | { /* "default" - for any srcSize > 256 KB */ |
4121 | | /* W, C, H, S, L, TL, strat */ |
4122 | | { 19, 12, 13, 1, 6, 1, ZSTD_fast }, /* base for negative levels */ |
4123 | | { 19, 13, 14, 1, 7, 0, ZSTD_fast }, /* level 1 */ |
4124 | | { 20, 15, 16, 1, 6, 0, ZSTD_fast }, /* level 2 */ |
4125 | | { 21, 16, 17, 1, 5, 0, ZSTD_dfast }, /* level 3 */ |
4126 | | { 21, 18, 18, 1, 5, 0, ZSTD_dfast }, /* level 4 */ |
4127 | | { 21, 18, 19, 2, 5, 2, ZSTD_greedy }, /* level 5 */ |
4128 | | { 21, 19, 19, 3, 5, 4, ZSTD_greedy }, /* level 6 */ |
4129 | | { 21, 19, 19, 3, 5, 8, ZSTD_lazy }, /* level 7 */ |
4130 | | { 21, 19, 19, 3, 5, 16, ZSTD_lazy2 }, /* level 8 */ |
4131 | | { 21, 19, 20, 4, 5, 16, ZSTD_lazy2 }, /* level 9 */ |
4132 | | { 22, 20, 21, 4, 5, 16, ZSTD_lazy2 }, /* level 10 */ |
4133 | | { 22, 21, 22, 4, 5, 16, ZSTD_lazy2 }, /* level 11 */ |
4134 | | { 22, 21, 22, 5, 5, 16, ZSTD_lazy2 }, /* level 12 */ |
4135 | | { 22, 21, 22, 5, 5, 32, ZSTD_btlazy2 }, /* level 13 */ |
4136 | | { 22, 22, 23, 5, 5, 32, ZSTD_btlazy2 }, /* level 14 */ |
4137 | | { 22, 23, 23, 6, 5, 32, ZSTD_btlazy2 }, /* level 15 */ |
4138 | | { 22, 22, 22, 5, 5, 48, ZSTD_btopt }, /* level 16 */ |
4139 | | { 23, 23, 22, 5, 4, 64, ZSTD_btopt }, /* level 17 */ |
4140 | | { 23, 23, 22, 6, 3, 64, ZSTD_btultra }, /* level 18 */ |
4141 | | { 23, 24, 22, 7, 3,256, ZSTD_btultra2}, /* level 19 */ |
4142 | | { 25, 25, 23, 7, 3,256, ZSTD_btultra2}, /* level 20 */ |
4143 | | { 26, 26, 24, 7, 3,512, ZSTD_btultra2}, /* level 21 */ |
4144 | | { 27, 27, 25, 9, 3,999, ZSTD_btultra2}, /* level 22 */ |
4145 | | }, |
4146 | | { /* for srcSize <= 256 KB */ |
4147 | | /* W, C, H, S, L, T, strat */ |
4148 | | { 18, 12, 13, 1, 5, 1, ZSTD_fast }, /* base for negative levels */ |
4149 | | { 18, 13, 14, 1, 6, 0, ZSTD_fast }, /* level 1 */ |
4150 | | { 18, 14, 14, 1, 5, 0, ZSTD_dfast }, /* level 2 */ |
4151 | | { 18, 16, 16, 1, 4, 0, ZSTD_dfast }, /* level 3 */ |
4152 | | { 18, 16, 17, 2, 5, 2, ZSTD_greedy }, /* level 4.*/ |
4153 | | { 18, 18, 18, 3, 5, 2, ZSTD_greedy }, /* level 5.*/ |
4154 | | { 18, 18, 19, 3, 5, 4, ZSTD_lazy }, /* level 6.*/ |
4155 | | { 18, 18, 19, 4, 4, 4, ZSTD_lazy }, /* level 7 */ |
4156 | | { 18, 18, 19, 4, 4, 8, ZSTD_lazy2 }, /* level 8 */ |
4157 | | { 18, 18, 19, 5, 4, 8, ZSTD_lazy2 }, /* level 9 */ |
4158 | | { 18, 18, 19, 6, 4, 8, ZSTD_lazy2 }, /* level 10 */ |
4159 | | { 18, 18, 19, 5, 4, 12, ZSTD_btlazy2 }, /* level 11.*/ |
4160 | | { 18, 19, 19, 7, 4, 12, ZSTD_btlazy2 }, /* level 12.*/ |
4161 | | { 18, 18, 19, 4, 4, 16, ZSTD_btopt }, /* level 13 */ |
4162 | | { 18, 18, 19, 4, 3, 32, ZSTD_btopt }, /* level 14.*/ |
4163 | | { 18, 18, 19, 6, 3,128, ZSTD_btopt }, /* level 15.*/ |
4164 | | { 18, 19, 19, 6, 3,128, ZSTD_btultra }, /* level 16.*/ |
4165 | | { 18, 19, 19, 8, 3,256, ZSTD_btultra }, /* level 17.*/ |
4166 | | { 18, 19, 19, 6, 3,128, ZSTD_btultra2}, /* level 18.*/ |
4167 | | { 18, 19, 19, 8, 3,256, ZSTD_btultra2}, /* level 19.*/ |
4168 | | { 18, 19, 19, 10, 3,512, ZSTD_btultra2}, /* level 20.*/ |
4169 | | { 18, 19, 19, 12, 3,512, ZSTD_btultra2}, /* level 21.*/ |
4170 | | { 18, 19, 19, 13, 3,999, ZSTD_btultra2}, /* level 22.*/ |
4171 | | }, |
4172 | | { /* for srcSize <= 128 KB */ |
4173 | | /* W, C, H, S, L, T, strat */ |
4174 | | { 17, 12, 12, 1, 5, 1, ZSTD_fast }, /* base for negative levels */ |
4175 | | { 17, 12, 13, 1, 6, 0, ZSTD_fast }, /* level 1 */ |
4176 | | { 17, 13, 15, 1, 5, 0, ZSTD_fast }, /* level 2 */ |
4177 | | { 17, 15, 16, 2, 5, 0, ZSTD_dfast }, /* level 3 */ |
4178 | | { 17, 17, 17, 2, 4, 0, ZSTD_dfast }, /* level 4 */ |
4179 | | { 17, 16, 17, 3, 4, 2, ZSTD_greedy }, /* level 5 */ |
4180 | | { 17, 17, 17, 3, 4, 4, ZSTD_lazy }, /* level 6 */ |
4181 | | { 17, 17, 17, 3, 4, 8, ZSTD_lazy2 }, /* level 7 */ |
4182 | | { 17, 17, 17, 4, 4, 8, ZSTD_lazy2 }, /* level 8 */ |
4183 | | { 17, 17, 17, 5, 4, 8, ZSTD_lazy2 }, /* level 9 */ |
4184 | | { 17, 17, 17, 6, 4, 8, ZSTD_lazy2 }, /* level 10 */ |
4185 | | { 17, 17, 17, 5, 4, 8, ZSTD_btlazy2 }, /* level 11 */ |
4186 | | { 17, 18, 17, 7, 4, 12, ZSTD_btlazy2 }, /* level 12 */ |
4187 | | { 17, 18, 17, 3, 4, 12, ZSTD_btopt }, /* level 13.*/ |
4188 | | { 17, 18, 17, 4, 3, 32, ZSTD_btopt }, /* level 14.*/ |
4189 | | { 17, 18, 17, 6, 3,256, ZSTD_btopt }, /* level 15.*/ |
4190 | | { 17, 18, 17, 6, 3,128, ZSTD_btultra }, /* level 16.*/ |
4191 | | { 17, 18, 17, 8, 3,256, ZSTD_btultra }, /* level 17.*/ |
4192 | | { 17, 18, 17, 10, 3,512, ZSTD_btultra }, /* level 18.*/ |
4193 | | { 17, 18, 17, 5, 3,256, ZSTD_btultra2}, /* level 19.*/ |
4194 | | { 17, 18, 17, 7, 3,512, ZSTD_btultra2}, /* level 20.*/ |
4195 | | { 17, 18, 17, 9, 3,512, ZSTD_btultra2}, /* level 21.*/ |
4196 | | { 17, 18, 17, 11, 3,999, ZSTD_btultra2}, /* level 22.*/ |
4197 | | }, |
4198 | | { /* for srcSize <= 16 KB */ |
4199 | | /* W, C, H, S, L, T, strat */ |
4200 | | { 14, 12, 13, 1, 5, 1, ZSTD_fast }, /* base for negative levels */ |
4201 | | { 14, 14, 15, 1, 5, 0, ZSTD_fast }, /* level 1 */ |
4202 | | { 14, 14, 15, 1, 4, 0, ZSTD_fast }, /* level 2 */ |
4203 | | { 14, 14, 15, 2, 4, 0, ZSTD_dfast }, /* level 3 */ |
4204 | | { 14, 14, 14, 4, 4, 2, ZSTD_greedy }, /* level 4 */ |
4205 | | { 14, 14, 14, 3, 4, 4, ZSTD_lazy }, /* level 5.*/ |
4206 | | { 14, 14, 14, 4, 4, 8, ZSTD_lazy2 }, /* level 6 */ |
4207 | | { 14, 14, 14, 6, 4, 8, ZSTD_lazy2 }, /* level 7 */ |
4208 | | { 14, 14, 14, 8, 4, 8, ZSTD_lazy2 }, /* level 8.*/ |
4209 | | { 14, 15, 14, 5, 4, 8, ZSTD_btlazy2 }, /* level 9.*/ |
4210 | | { 14, 15, 14, 9, 4, 8, ZSTD_btlazy2 }, /* level 10.*/ |
4211 | | { 14, 15, 14, 3, 4, 12, ZSTD_btopt }, /* level 11.*/ |
4212 | | { 14, 15, 14, 4, 3, 24, ZSTD_btopt }, /* level 12.*/ |
4213 | | { 14, 15, 14, 5, 3, 32, ZSTD_btultra }, /* level 13.*/ |
4214 | | { 14, 15, 15, 6, 3, 64, ZSTD_btultra }, /* level 14.*/ |
4215 | | { 14, 15, 15, 7, 3,256, ZSTD_btultra }, /* level 15.*/ |
4216 | | { 14, 15, 15, 5, 3, 48, ZSTD_btultra2}, /* level 16.*/ |
4217 | | { 14, 15, 15, 6, 3,128, ZSTD_btultra2}, /* level 17.*/ |
4218 | | { 14, 15, 15, 7, 3,256, ZSTD_btultra2}, /* level 18.*/ |
4219 | | { 14, 15, 15, 8, 3,256, ZSTD_btultra2}, /* level 19.*/ |
4220 | | { 14, 15, 15, 8, 3,512, ZSTD_btultra2}, /* level 20.*/ |
4221 | | { 14, 15, 15, 9, 3,512, ZSTD_btultra2}, /* level 21.*/ |
4222 | | { 14, 15, 15, 10, 3,999, ZSTD_btultra2}, /* level 22.*/ |
4223 | | }, |
4224 | | }; |
4225 | | |
4226 | | /*! ZSTD_getCParams_internal() : |
4227 | | * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize. |
4228 | | * Note: srcSizeHint 0 means 0, use ZSTD_CONTENTSIZE_UNKNOWN for unknown. |
4229 | | * Use dictSize == 0 for unknown or unused. */ |
4230 | | static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) |
4231 | 0 | { |
4232 | 0 | int const unknown = srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN; |
4233 | 0 | size_t const addedSize = unknown && dictSize > 0 ? 500 : 0; |
4234 | 0 | U64 const rSize = unknown && dictSize == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : srcSizeHint+dictSize+addedSize; |
4235 | 0 | U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB); |
4236 | 0 | int row = compressionLevel; |
4237 | 0 | DEBUGLOG(5, "ZSTD_getCParams_internal (cLevel=%i)", compressionLevel); |
4238 | 0 | if (compressionLevel == 0) row = ZSTD_CLEVEL_DEFAULT; /* 0 == default */ |
4239 | 0 | if (compressionLevel < 0) row = 0; /* entry 0 is baseline for fast mode */ |
4240 | 0 | if (compressionLevel > ZSTD_MAX_CLEVEL) row = ZSTD_MAX_CLEVEL; |
4241 | 0 | { ZSTD_compressionParameters cp = ZSTD_defaultCParameters[tableID][row]; |
4242 | 0 | if (compressionLevel < 0) cp.targetLength = (unsigned)(-compressionLevel); /* acceleration factor */ |
4243 | | /* refine parameters based on srcSize & dictSize */ |
4244 | 0 | return ZSTD_adjustCParams_internal(cp, srcSizeHint, dictSize); |
4245 | 0 | } |
4246 | 0 | } |
4247 | | |
4248 | | /*! ZSTD_getCParams() : |
4249 | | * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize. |
4250 | | * Size values are optional, provide 0 if not known or unused */ |
4251 | | ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) |
4252 | 0 | { |
4253 | 0 | if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN; |
4254 | 0 | return ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize); |
4255 | 0 | } |
4256 | | |
4257 | | /*! ZSTD_getParams() : |
4258 | | * same idea as ZSTD_getCParams() |
4259 | | * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`). |
4260 | | * Fields of `ZSTD_frameParameters` are set to default values */ |
4261 | 0 | static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) { |
4262 | 0 | ZSTD_parameters params; |
4263 | 0 | ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize); |
4264 | 0 | DEBUGLOG(5, "ZSTD_getParams (cLevel=%i)", compressionLevel); |
4265 | 0 | memset(¶ms, 0, sizeof(params)); |
4266 | 0 | params.cParams = cParams; |
4267 | 0 | params.fParams.contentSizeFlag = 1; |
4268 | 0 | return params; |
4269 | 0 | } |
4270 | | |
4271 | | /*! ZSTD_getParams() : |
4272 | | * same idea as ZSTD_getCParams() |
4273 | | * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`). |
4274 | | * Fields of `ZSTD_frameParameters` are set to default values */ |
4275 | 0 | ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) { |
4276 | 0 | if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN; |
4277 | 0 | return ZSTD_getParams_internal(compressionLevel, srcSizeHint, dictSize); |
4278 | 0 | } |
4279 | | |
4280 | | } |