1 /* SPDX-License-Identifier: (GPL-2.0 or BSD-3-Clause-Clear) */
3 * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
10 /* ====== Dependency ======*/
11 #include <linux/types.h> /* size_t */
14 /*-*****************************************************************************
17 * zstd, short for Zstandard, is a fast lossless compression algorithm,
18 * targeting real-time compression scenarios at zlib-level and better
19 * compression ratios. The zstd compression library provides in-memory
20 * compression and decompression functions. The library supports compression
21 * levels from 1 up to ZSTD_maxCLevel() which is 22. Levels >= 20, labeled
22 * ultra, should be used with caution, as they require more memory.
23 * Compression can be done in:
24 * - a single step, reusing a context (described as Explicit memory management)
25 * - unbounded multiple steps (described as Streaming compression)
26 * The compression ratio achievable on small data can be highly improved using
27 * compression with a dictionary in:
28 * - a single step (described as Simple dictionary API)
29 * - a single step, reusing a dictionary (described as Fast dictionary API)
30 ******************************************************************************/
32 /*====== Helper functions ======*/
35 * enum ZSTD_ErrorCode - zstd error codes
37 * Functions that return size_t can be checked for errors using ZSTD_isError()
38 * and the ZSTD_ErrorCode can be extracted using ZSTD_getErrorCode().
43 ZSTD_error_prefix_unknown,
44 ZSTD_error_version_unsupported,
45 ZSTD_error_parameter_unknown,
46 ZSTD_error_frameParameter_unsupported,
47 ZSTD_error_frameParameter_unsupportedBy32bits,
48 ZSTD_error_frameParameter_windowTooLarge,
49 ZSTD_error_compressionParameter_unsupported,
50 ZSTD_error_init_missing,
51 ZSTD_error_memory_allocation,
52 ZSTD_error_stage_wrong,
53 ZSTD_error_dstSize_tooSmall,
54 ZSTD_error_srcSize_wrong,
55 ZSTD_error_corruption_detected,
56 ZSTD_error_checksum_wrong,
57 ZSTD_error_tableLog_tooLarge,
58 ZSTD_error_maxSymbolValue_tooLarge,
59 ZSTD_error_maxSymbolValue_tooSmall,
60 ZSTD_error_dictionary_corrupted,
61 ZSTD_error_dictionary_wrong,
62 ZSTD_error_dictionaryCreation_failed,
67 * ZSTD_maxCLevel() - maximum compression level available
69 * Return: Maximum compression level available.
71 int ZSTD_maxCLevel(void);
73 * ZSTD_compressBound() - maximum compressed size in worst case scenario
74 * @srcSize: The size of the data to compress.
76 * Return: The maximum compressed size in the worst case scenario.
78 size_t ZSTD_compressBound(size_t srcSize);
80 * ZSTD_isError() - tells if a size_t function result is an error code
81 * @code: The function result to check for error.
83 * Return: Non-zero iff the code is an error.
85 static __attribute__((unused)) unsigned int ZSTD_isError(size_t code)
87 return code > (size_t)-ZSTD_error_maxCode;
90 * ZSTD_getErrorCode() - translates an error function result to a ZSTD_ErrorCode
91 * @functionResult: The result of a function for which ZSTD_isError() is true.
93 * Return: The ZSTD_ErrorCode corresponding to the functionResult or 0
94 * if the functionResult isn't an error.
96 static __attribute__((unused)) ZSTD_ErrorCode ZSTD_getErrorCode(
97 size_t functionResult)
99 if (!ZSTD_isError(functionResult))
100 return (ZSTD_ErrorCode)0;
101 return (ZSTD_ErrorCode)(0 - functionResult);
105 * enum ZSTD_strategy - zstd compression search strategy
107 * From faster to stronger.
121 * struct ZSTD_compressionParameters - zstd compression parameters
122 * @windowLog: Log of the largest match distance. Larger means more
123 * compression, and more memory needed during decompression.
124 * @chainLog: Fully searched segment. Larger means more compression, slower,
125 * and more memory (useless for fast).
126 * @hashLog: Dispatch table. Larger means more compression,
127 * slower, and more memory.
128 * @searchLog: Number of searches. Larger means more compression and slower.
129 * @searchLength: Match length searched. Larger means faster decompression,
130 * sometimes less compression.
131 * @targetLength: Acceptable match size for optimal parser (only). Larger means
132 * more compression, and slower.
133 * @strategy: The zstd compression strategy.
136 unsigned int windowLog;
137 unsigned int chainLog;
138 unsigned int hashLog;
139 unsigned int searchLog;
140 unsigned int searchLength;
141 unsigned int targetLength;
142 ZSTD_strategy strategy;
143 } ZSTD_compressionParameters;
146 * struct ZSTD_frameParameters - zstd frame parameters
147 * @contentSizeFlag: Controls whether content size will be present in the frame
148 * header (when known).
149 * @checksumFlag: Controls whether a 32-bit checksum is generated at the end
150 * of the frame for error detection.
151 * @noDictIDFlag: Controls whether dictID will be saved into the frame header
152 * when using dictionary compression.
154 * The default value is all fields set to 0.
157 unsigned int contentSizeFlag;
158 unsigned int checksumFlag;
159 unsigned int noDictIDFlag;
160 } ZSTD_frameParameters;
163 * struct ZSTD_parameters - zstd parameters
164 * @cParams: The compression parameters.
165 * @fParams: The frame parameters.
168 ZSTD_compressionParameters cParams;
169 ZSTD_frameParameters fParams;
173 * ZSTD_getCParams() - returns ZSTD_compressionParameters for selected level
174 * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
175 * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
176 * @dictSize: The dictionary size or 0 if a dictionary isn't being used.
178 * Return: The selected ZSTD_compressionParameters.
180 ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel,
181 unsigned long long estimatedSrcSize, size_t dictSize);
184 * ZSTD_getParams() - returns ZSTD_parameters for selected level
185 * @compressionLevel: The compression level from 1 to ZSTD_maxCLevel().
186 * @estimatedSrcSize: The estimated source size to compress or 0 if unknown.
187 * @dictSize: The dictionary size or 0 if a dictionary isn't being used.
189 * The same as ZSTD_getCParams() except also selects the default frame
190 * parameters (all zero).
192 * Return: The selected ZSTD_parameters.
194 ZSTD_parameters ZSTD_getParams(int compressionLevel,
195 unsigned long long estimatedSrcSize, size_t dictSize);
197 /*-*************************************
198 * Explicit memory management
199 **************************************/
202 * ZSTD_CCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_CCtx
203 * @cParams: The compression parameters to be used for compression.
205 * If multiple compression parameters might be used, the caller must call
206 * ZSTD_CCtxWorkspaceBound() for each set of parameters and use the maximum
209 * Return: A lower bound on the size of the workspace that is passed to
212 size_t ZSTD_CCtxWorkspaceBound(ZSTD_compressionParameters cParams);
215 * struct ZSTD_CCtx - the zstd compression context
217 * When compressing many times it is recommended to allocate a context just once
218 * and reuse it for each successive compression operation.
220 typedef struct ZSTD_CCtx_s ZSTD_CCtx;
222 * ZSTD_initCCtx() - initialize a zstd compression context
223 * @workspace: The workspace to emplace the context into. It must outlive
224 * the returned context.
225 * @workspaceSize: The size of workspace. Use ZSTD_CCtxWorkspaceBound() to
226 * determine how large the workspace must be.
228 * Return: A compression context emplaced into workspace.
230 ZSTD_CCtx *ZSTD_initCCtx(void *workspace, size_t workspaceSize);
233 * ZSTD_compressCCtx() - compress src into dst
234 * @ctx: The context. Must have been initialized with a workspace at
235 * least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
236 * @dst: The buffer to compress src into.
237 * @dstCapacity: The size of the destination buffer. May be any size, but
238 * ZSTD_compressBound(srcSize) is guaranteed to be large enough.
239 * @src: The data to compress.
240 * @srcSize: The size of the data to compress.
241 * @params: The parameters to use for compression. See ZSTD_getParams().
243 * Return: The compressed size or an error, which can be checked using
246 size_t ZSTD_compressCCtx(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
247 const void *src, size_t srcSize, ZSTD_parameters params);
250 * ZSTD_DCtxWorkspaceBound() - amount of memory needed to initialize a ZSTD_DCtx
252 * Return: A lower bound on the size of the workspace that is passed to
255 size_t ZSTD_DCtxWorkspaceBound(void);
258 * struct ZSTD_DCtx - the zstd decompression context
260 * When decompressing many times it is recommended to allocate a context just
261 * once and reuse it for each successive decompression operation.
263 typedef struct ZSTD_DCtx_s ZSTD_DCtx;
265 * ZSTD_initDCtx() - initialize a zstd decompression context
266 * @workspace: The workspace to emplace the context into. It must outlive
267 * the returned context.
268 * @workspaceSize: The size of workspace. Use ZSTD_DCtxWorkspaceBound() to
269 * determine how large the workspace must be.
271 * Return: A decompression context emplaced into workspace.
273 ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize);
276 * ZSTD_decompressDCtx() - decompress zstd compressed src into dst
277 * @ctx: The decompression context.
278 * @dst: The buffer to decompress src into.
279 * @dstCapacity: The size of the destination buffer. Must be at least as large
280 * as the decompressed size. If the caller cannot upper bound the
281 * decompressed size, then it's better to use the streaming API.
282 * @src: The zstd compressed data to decompress. Multiple concatenated
283 * frames and skippable frames are allowed.
284 * @srcSize: The exact size of the data to decompress.
286 * Return: The decompressed size or an error, which can be checked using
289 size_t ZSTD_decompressDCtx(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
290 const void *src, size_t srcSize);
292 /*-************************
293 * Simple dictionary API
294 **************************/
297 * ZSTD_compress_usingDict() - compress src into dst using a dictionary
298 * @ctx: The context. Must have been initialized with a workspace at
299 * least as large as ZSTD_CCtxWorkspaceBound(params.cParams).
300 * @dst: The buffer to compress src into.
301 * @dstCapacity: The size of the destination buffer. May be any size, but
302 * ZSTD_compressBound(srcSize) is guaranteed to be large enough.
303 * @src: The data to compress.
304 * @srcSize: The size of the data to compress.
305 * @dict: The dictionary to use for compression.
306 * @dictSize: The size of the dictionary.
307 * @params: The parameters to use for compression. See ZSTD_getParams().
309 * Compression using a predefined dictionary. The same dictionary must be used
310 * during decompression.
312 * Return: The compressed size or an error, which can be checked using
315 size_t ZSTD_compress_usingDict(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity,
316 const void *src, size_t srcSize, const void *dict, size_t dictSize,
317 ZSTD_parameters params);
320 * ZSTD_decompress_usingDict() - decompress src into dst using a dictionary
321 * @ctx: The decompression context.
322 * @dst: The buffer to decompress src into.
323 * @dstCapacity: The size of the destination buffer. Must be at least as large
324 * as the decompressed size. If the caller cannot upper bound the
325 * decompressed size, then it's better to use the streaming API.
326 * @src: The zstd compressed data to decompress. Multiple concatenated
327 * frames and skippable frames are allowed.
328 * @srcSize: The exact size of the data to decompress.
329 * @dict: The dictionary to use for decompression. The same dictionary
330 * must've been used to compress the data.
331 * @dictSize: The size of the dictionary.
333 * Return: The decompressed size or an error, which can be checked using
336 size_t ZSTD_decompress_usingDict(ZSTD_DCtx *ctx, void *dst, size_t dstCapacity,
337 const void *src, size_t srcSize, const void *dict, size_t dictSize);
339 /*-**************************
340 * Fast dictionary API
341 ***************************/
344 * ZSTD_CDictWorkspaceBound() - memory needed to initialize a ZSTD_CDict
345 * @cParams: The compression parameters to be used for compression.
347 * Return: A lower bound on the size of the workspace that is passed to
350 size_t ZSTD_CDictWorkspaceBound(ZSTD_compressionParameters cParams);
353 * struct ZSTD_CDict - a digested dictionary to be used for compression
355 typedef struct ZSTD_CDict_s ZSTD_CDict;
358 * ZSTD_initCDict() - initialize a digested dictionary for compression
359 * @dictBuffer: The dictionary to digest. The buffer is referenced by the
360 * ZSTD_CDict so it must outlive the returned ZSTD_CDict.
361 * @dictSize: The size of the dictionary.
362 * @params: The parameters to use for compression. See ZSTD_getParams().
363 * @workspace: The workspace. It must outlive the returned ZSTD_CDict.
364 * @workspaceSize: The workspace size. Must be at least
365 * ZSTD_CDictWorkspaceBound(params.cParams).
367 * When compressing multiple messages / blocks with the same dictionary it is
368 * recommended to load it just once. The ZSTD_CDict merely references the
369 * dictBuffer, so it must outlive the returned ZSTD_CDict.
371 * Return: The digested dictionary emplaced into workspace.
373 ZSTD_CDict *ZSTD_initCDict(const void *dictBuffer, size_t dictSize,
374 ZSTD_parameters params, void *workspace, size_t workspaceSize);
377 * ZSTD_compress_usingCDict() - compress src into dst using a ZSTD_CDict
378 * @ctx: The context. Must have been initialized with a workspace at
379 * least as large as ZSTD_CCtxWorkspaceBound(cParams) where
380 * cParams are the compression parameters used to initialize the
382 * @dst: The buffer to compress src into.
383 * @dstCapacity: The size of the destination buffer. May be any size, but
384 * ZSTD_compressBound(srcSize) is guaranteed to be large enough.
385 * @src: The data to compress.
386 * @srcSize: The size of the data to compress.
387 * @cdict: The digested dictionary to use for compression.
388 * @params: The parameters to use for compression. See ZSTD_getParams().
390 * Compression using a digested dictionary. The same dictionary must be used
391 * during decompression.
393 * Return: The compressed size or an error, which can be checked using
396 size_t ZSTD_compress_usingCDict(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
397 const void *src, size_t srcSize, const ZSTD_CDict *cdict);
401 * ZSTD_DDictWorkspaceBound() - memory needed to initialize a ZSTD_DDict
403 * Return: A lower bound on the size of the workspace that is passed to
406 size_t ZSTD_DDictWorkspaceBound(void);
409 * struct ZSTD_DDict - a digested dictionary to be used for decompression
411 typedef struct ZSTD_DDict_s ZSTD_DDict;
414 * ZSTD_initDDict() - initialize a digested dictionary for decompression
415 * @dictBuffer: The dictionary to digest. The buffer is referenced by the
416 * ZSTD_DDict so it must outlive the returned ZSTD_DDict.
417 * @dictSize: The size of the dictionary.
418 * @workspace: The workspace. It must outlive the returned ZSTD_DDict.
419 * @workspaceSize: The workspace size. Must be at least
420 * ZSTD_DDictWorkspaceBound().
422 * When decompressing multiple messages / blocks with the same dictionary it is
423 * recommended to load it just once. The ZSTD_DDict merely references the
424 * dictBuffer, so it must outlive the returned ZSTD_DDict.
426 * Return: The digested dictionary emplaced into workspace.
428 ZSTD_DDict *ZSTD_initDDict(const void *dictBuffer, size_t dictSize,
429 void *workspace, size_t workspaceSize);
432 * ZSTD_decompress_usingDDict() - decompress src into dst using a ZSTD_DDict
433 * @ctx: The decompression context.
434 * @dst: The buffer to decompress src into.
435 * @dstCapacity: The size of the destination buffer. Must be at least as large
436 * as the decompressed size. If the caller cannot upper bound the
437 * decompressed size, then it's better to use the streaming API.
438 * @src: The zstd compressed data to decompress. Multiple concatenated
439 * frames and skippable frames are allowed.
440 * @srcSize: The exact size of the data to decompress.
441 * @ddict: The digested dictionary to use for decompression. The same
442 * dictionary must've been used to compress the data.
444 * Return: The decompressed size or an error, which can be checked using
447 size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst,
448 size_t dstCapacity, const void *src, size_t srcSize,
449 const ZSTD_DDict *ddict);
452 /*-**************************
454 ***************************/
457 * struct ZSTD_inBuffer - input buffer for streaming
458 * @src: Start of the input buffer.
459 * @size: Size of the input buffer.
460 * @pos: Position where reading stopped. Will be updated.
461 * Necessarily 0 <= pos <= size.
463 typedef struct ZSTD_inBuffer_s {
470 * struct ZSTD_outBuffer - output buffer for streaming
471 * @dst: Start of the output buffer.
472 * @size: Size of the output buffer.
473 * @pos: Position where writing stopped. Will be updated.
474 * Necessarily 0 <= pos <= size.
476 typedef struct ZSTD_outBuffer_s {
484 /*-*****************************************************************************
485 * Streaming compression - HowTo
487 * A ZSTD_CStream object is required to track streaming operation.
488 * Use ZSTD_initCStream() to initialize a ZSTD_CStream object.
489 * ZSTD_CStream objects can be reused multiple times on consecutive compression
490 * operations. It is recommended to re-use ZSTD_CStream in situations where many
491 * streaming operations will be achieved consecutively. Use one separate
492 * ZSTD_CStream per thread for parallel execution.
494 * Use ZSTD_compressStream() repetitively to consume input stream.
495 * The function will automatically update both `pos` fields.
496 * Note that it may not consume the entire input, in which case `pos < size`,
497 * and it's up to the caller to present again remaining data.
498 * It returns a hint for the preferred number of bytes to use as an input for
499 * the next function call.
501 * At any moment, it's possible to flush whatever data remains within internal
502 * buffer, using ZSTD_flushStream(). `output->pos` will be updated. There might
503 * still be some content left within the internal buffer if `output->size` is
504 * too small. It returns the number of bytes left in the internal buffer and
505 * must be called until it returns 0.
507 * ZSTD_endStream() instructs to finish a frame. It will perform a flush and
508 * write frame epilogue. The epilogue is required for decoders to consider a
509 * frame completed. Similar to ZSTD_flushStream(), it may not be able to flush
510 * the full content if `output->size` is too small. In which case, call again
511 * ZSTD_endStream() to complete the flush. It returns the number of bytes left
512 * in the internal buffer and must be called until it returns 0.
513 ******************************************************************************/
516 * ZSTD_CStreamWorkspaceBound() - memory needed to initialize a ZSTD_CStream
517 * @cParams: The compression parameters to be used for compression.
519 * Return: A lower bound on the size of the workspace that is passed to
520 * ZSTD_initCStream() and ZSTD_initCStream_usingCDict().
522 size_t ZSTD_CStreamWorkspaceBound(ZSTD_compressionParameters cParams);
525 * struct ZSTD_CStream - the zstd streaming compression context
527 typedef struct ZSTD_CStream_s ZSTD_CStream;
529 /*===== ZSTD_CStream management functions =====*/
531 * ZSTD_initCStream() - initialize a zstd streaming compression context
532 * @params: The zstd compression parameters.
533 * @pledgedSrcSize: If params.fParams.contentSizeFlag == 1 then the caller must
534 * pass the source size (zero means empty source). Otherwise,
535 * the caller may optionally pass the source size, or zero if
537 * @workspace: The workspace to emplace the context into. It must outlive
538 * the returned context.
539 * @workspaceSize: The size of workspace.
540 * Use ZSTD_CStreamWorkspaceBound(params.cParams) to determine
541 * how large the workspace must be.
543 * Return: The zstd streaming compression context.
545 ZSTD_CStream *ZSTD_initCStream(ZSTD_parameters params,
546 unsigned long long pledgedSrcSize, void *workspace,
547 size_t workspaceSize);
550 * ZSTD_initCStream_usingCDict() - initialize a streaming compression context
551 * @cdict: The digested dictionary to use for compression.
552 * @pledgedSrcSize: Optionally the source size, or zero if unknown.
553 * @workspace: The workspace to emplace the context into. It must outlive
554 * the returned context.
555 * @workspaceSize: The size of workspace. Call ZSTD_CStreamWorkspaceBound()
556 * with the cParams used to initialize the cdict to determine
557 * how large the workspace must be.
559 * Return: The zstd streaming compression context.
561 ZSTD_CStream *ZSTD_initCStream_usingCDict(const ZSTD_CDict *cdict,
562 unsigned long long pledgedSrcSize, void *workspace,
563 size_t workspaceSize);
565 /*===== Streaming compression functions =====*/
567 * ZSTD_resetCStream() - reset the context using parameters from creation
568 * @zcs: The zstd streaming compression context to reset.
569 * @pledgedSrcSize: Optionally the source size, or zero if unknown.
571 * Resets the context using the parameters from creation. Skips dictionary
572 * loading, since it can be reused. If `pledgedSrcSize` is non-zero the frame
573 * content size is always written into the frame header.
575 * Return: Zero or an error, which can be checked using ZSTD_isError().
577 size_t ZSTD_resetCStream(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize);
579 * ZSTD_compressStream() - streaming compress some of input into output
580 * @zcs: The zstd streaming compression context.
581 * @output: Destination buffer. `output->pos` is updated to indicate how much
582 * compressed data was written.
583 * @input: Source buffer. `input->pos` is updated to indicate how much data was
584 * read. Note that it may not consume the entire input, in which case
585 * `input->pos < input->size`, and it's up to the caller to present
586 * remaining data again.
588 * The `input` and `output` buffers may be any size. Guaranteed to make some
589 * forward progress if `input` and `output` are not empty.
591 * Return: A hint for the number of bytes to use as the input for the next
592 * function call or an error, which can be checked using
595 size_t ZSTD_compressStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output,
596 ZSTD_inBuffer *input);
598 * ZSTD_flushStream() - flush internal buffers into output
599 * @zcs: The zstd streaming compression context.
600 * @output: Destination buffer. `output->pos` is updated to indicate how much
601 * compressed data was written.
603 * ZSTD_flushStream() must be called until it returns 0, meaning all the data
604 * has been flushed. Since ZSTD_flushStream() causes a block to be ended,
605 * calling it too often will degrade the compression ratio.
607 * Return: The number of bytes still present within internal buffers or an
608 * error, which can be checked using ZSTD_isError().
610 size_t ZSTD_flushStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
612 * ZSTD_endStream() - flush internal buffers into output and end the frame
613 * @zcs: The zstd streaming compression context.
614 * @output: Destination buffer. `output->pos` is updated to indicate how much
615 * compressed data was written.
617 * ZSTD_endStream() must be called until it returns 0, meaning all the data has
618 * been flushed and the frame epilogue has been written.
620 * Return: The number of bytes still present within internal buffers or an
621 * error, which can be checked using ZSTD_isError().
623 size_t ZSTD_endStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output);
626 * ZSTD_CStreamInSize() - recommended size for the input buffer
628 * Return: The recommended size for the input buffer.
630 size_t ZSTD_CStreamInSize(void);
632 * ZSTD_CStreamOutSize() - recommended size for the output buffer
634 * When the output buffer is at least this large, it is guaranteed to be large
635 * enough to flush at least one complete compressed block.
637 * Return: The recommended size for the output buffer.
639 size_t ZSTD_CStreamOutSize(void);
643 /*-*****************************************************************************
644 * Streaming decompression - HowTo
646 * A ZSTD_DStream object is required to track streaming operations.
647 * Use ZSTD_initDStream() to initialize a ZSTD_DStream object.
648 * ZSTD_DStream objects can be re-used multiple times.
650 * Use ZSTD_decompressStream() repetitively to consume your input.
651 * The function will update both `pos` fields.
652 * If `input->pos < input->size`, some input has not been consumed.
653 * It's up to the caller to present again remaining data.
654 * If `output->pos < output->size`, decoder has flushed everything it could.
655 * Returns 0 iff a frame is completely decoded and fully flushed.
656 * Otherwise it returns a suggested next input size that will never load more
657 * than the current frame.
658 ******************************************************************************/
661 * ZSTD_DStreamWorkspaceBound() - memory needed to initialize a ZSTD_DStream
662 * @maxWindowSize: The maximum window size allowed for compressed frames.
664 * Return: A lower bound on the size of the workspace that is passed to
665 * ZSTD_initDStream() and ZSTD_initDStream_usingDDict().
667 size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize);
670 * struct ZSTD_DStream - the zstd streaming decompression context
672 typedef struct ZSTD_DStream_s ZSTD_DStream;
673 /*===== ZSTD_DStream management functions =====*/
675 * ZSTD_initDStream() - initialize a zstd streaming decompression context
676 * @maxWindowSize: The maximum window size allowed for compressed frames.
677 * @workspace: The workspace to emplace the context into. It must outlive
678 * the returned context.
679 * @workspaceSize: The size of workspace.
680 * Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
681 * how large the workspace must be.
683 * Return: The zstd streaming decompression context.
685 ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace,
686 size_t workspaceSize);
688 * ZSTD_initDStream_usingDDict() - initialize streaming decompression context
689 * @maxWindowSize: The maximum window size allowed for compressed frames.
690 * @ddict: The digested dictionary to use for decompression.
691 * @workspace: The workspace to emplace the context into. It must outlive
692 * the returned context.
693 * @workspaceSize: The size of workspace.
694 * Use ZSTD_DStreamWorkspaceBound(maxWindowSize) to determine
695 * how large the workspace must be.
697 * Return: The zstd streaming decompression context.
699 ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize,
700 const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize);
702 /*===== Streaming decompression functions =====*/
704 * ZSTD_resetDStream() - reset the context using parameters from creation
705 * @zds: The zstd streaming decompression context to reset.
707 * Resets the context using the parameters from creation. Skips dictionary
708 * loading, since it can be reused.
710 * Return: Zero or an error, which can be checked using ZSTD_isError().
712 size_t ZSTD_resetDStream(ZSTD_DStream *zds);
714 * ZSTD_decompressStream() - streaming decompress some of input into output
715 * @zds: The zstd streaming decompression context.
716 * @output: Destination buffer. `output.pos` is updated to indicate how much
717 * decompressed data was written.
718 * @input: Source buffer. `input.pos` is updated to indicate how much data was
719 * read. Note that it may not consume the entire input, in which case
720 * `input.pos < input.size`, and it's up to the caller to present
721 * remaining data again.
723 * The `input` and `output` buffers may be any size. Guaranteed to make some
724 * forward progress if `input` and `output` are not empty.
725 * ZSTD_decompressStream() will not consume the last byte of the frame until
726 * the entire frame is flushed.
728 * Return: Returns 0 iff a frame is completely decoded and fully flushed.
729 * Otherwise returns a hint for the number of bytes to use as the input
730 * for the next function call or an error, which can be checked using
731 * ZSTD_isError(). The size hint will never load more than the frame.
733 size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output,
734 ZSTD_inBuffer *input);
737 * ZSTD_DStreamInSize() - recommended size for the input buffer
739 * Return: The recommended size for the input buffer.
741 size_t ZSTD_DStreamInSize(void);
743 * ZSTD_DStreamOutSize() - recommended size for the output buffer
745 * When the output buffer is at least this large, it is guaranteed to be large
746 * enough to flush at least one complete decompressed block.
748 * Return: The recommended size for the output buffer.
750 size_t ZSTD_DStreamOutSize(void);
753 /* --- Constants ---*/
754 #define ZSTD_MAGICNUMBER 0xFD2FB528 /* >= v0.8.0 */
755 #define ZSTD_MAGIC_SKIPPABLE_START 0x184D2A50U
757 #define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
758 #define ZSTD_CONTENTSIZE_ERROR (0ULL - 2)
760 #define ZSTD_WINDOWLOG_MAX_32 27
761 #define ZSTD_WINDOWLOG_MAX_64 27
762 #define ZSTD_WINDOWLOG_MAX \
763 ((unsigned int)(sizeof(size_t) == 4 \
764 ? ZSTD_WINDOWLOG_MAX_32 \
765 : ZSTD_WINDOWLOG_MAX_64))
766 #define ZSTD_WINDOWLOG_MIN 10
767 #define ZSTD_HASHLOG_MAX ZSTD_WINDOWLOG_MAX
768 #define ZSTD_HASHLOG_MIN 6
769 #define ZSTD_CHAINLOG_MAX (ZSTD_WINDOWLOG_MAX+1)
770 #define ZSTD_CHAINLOG_MIN ZSTD_HASHLOG_MIN
771 #define ZSTD_HASHLOG3_MAX 17
772 #define ZSTD_SEARCHLOG_MAX (ZSTD_WINDOWLOG_MAX-1)
773 #define ZSTD_SEARCHLOG_MIN 1
774 /* only for ZSTD_fast, other strategies are limited to 6 */
775 #define ZSTD_SEARCHLENGTH_MAX 7
776 /* only for ZSTD_btopt, other strategies are limited to 4 */
777 #define ZSTD_SEARCHLENGTH_MIN 3
778 #define ZSTD_TARGETLENGTH_MIN 4
779 #define ZSTD_TARGETLENGTH_MAX 999
781 /* for static allocation */
782 #define ZSTD_FRAMEHEADERSIZE_MAX 18
783 #define ZSTD_FRAMEHEADERSIZE_MIN 6
784 static const size_t ZSTD_frameHeaderSize_prefix = 5;
785 static const size_t ZSTD_frameHeaderSize_min = ZSTD_FRAMEHEADERSIZE_MIN;
786 static const size_t ZSTD_frameHeaderSize_max = ZSTD_FRAMEHEADERSIZE_MAX;
787 /* magic number + skippable frame length */
788 static const size_t ZSTD_skippableHeaderSize = 8;
791 /*-*************************************
792 * Compressed size functions
793 **************************************/
796 * ZSTD_findFrameCompressedSize() - returns the size of a compressed frame
797 * @src: Source buffer. It should point to the start of a zstd encoded frame
798 * or a skippable frame.
799 * @srcSize: The size of the source buffer. It must be at least as large as the
802 * Return: The compressed size of the frame pointed to by `src` or an error,
803 * which can be check with ZSTD_isError().
804 * Suitable to pass to ZSTD_decompress() or similar functions.
806 size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize);
808 /*-*************************************
809 * Decompressed size functions
810 **************************************/
812 * ZSTD_getFrameContentSize() - returns the content size in a zstd frame header
813 * @src: It should point to the start of a zstd encoded frame.
814 * @srcSize: The size of the source buffer. It must be at least as large as the
815 * frame header. `ZSTD_frameHeaderSize_max` is always large enough.
817 * Return: The frame content size stored in the frame header if known.
818 * `ZSTD_CONTENTSIZE_UNKNOWN` if the content size isn't stored in the
819 * frame header. `ZSTD_CONTENTSIZE_ERROR` on invalid input.
821 unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
824 * ZSTD_findDecompressedSize() - returns decompressed size of a series of frames
825 * @src: It should point to the start of a series of zstd encoded and/or
827 * @srcSize: The exact size of the series of frames.
829 * If any zstd encoded frame in the series doesn't have the frame content size
830 * set, `ZSTD_CONTENTSIZE_UNKNOWN` is returned. But frame content size is always
831 * set when using ZSTD_compress(). The decompressed size can be very large.
832 * If the source is untrusted, the decompressed size could be wrong or
833 * intentionally modified. Always ensure the result fits within the
834 * application's authorized limits. ZSTD_findDecompressedSize() handles multiple
835 * frames, and so it must traverse the input to read each frame header. This is
836 * efficient as most of the data is skipped, however it does mean that all frame
837 * data must be present and valid.
839 * Return: Decompressed size of all the data contained in the frames if known.
840 * `ZSTD_CONTENTSIZE_UNKNOWN` if the decompressed size is unknown.
841 * `ZSTD_CONTENTSIZE_ERROR` if an error occurred.
843 unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize);
845 /*-*************************************
846 * Advanced compression functions
847 **************************************/
849 * ZSTD_checkCParams() - ensure parameter values remain within authorized range
850 * @cParams: The zstd compression parameters.
852 * Return: Zero or an error, which can be checked using ZSTD_isError().
854 size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams);
857 * ZSTD_adjustCParams() - optimize parameters for a given srcSize and dictSize
858 * @srcSize: Optionally the estimated source size, or zero if unknown.
859 * @dictSize: Optionally the estimated dictionary size, or zero if unknown.
861 * Return: The optimized parameters.
863 ZSTD_compressionParameters ZSTD_adjustCParams(
864 ZSTD_compressionParameters cParams, unsigned long long srcSize,
867 /*--- Advanced decompression functions ---*/
870 * ZSTD_isFrame() - returns true iff the buffer starts with a valid frame
871 * @buffer: The source buffer to check.
872 * @size: The size of the source buffer, must be at least 4 bytes.
874 * Return: True iff the buffer starts with a zstd or skippable frame identifier.
876 unsigned int ZSTD_isFrame(const void *buffer, size_t size);
879 * ZSTD_getDictID_fromDict() - returns the dictionary id stored in a dictionary
880 * @dict: The dictionary buffer.
881 * @dictSize: The size of the dictionary buffer.
883 * Return: The dictionary id stored within the dictionary or 0 if the
884 * dictionary is not a zstd dictionary. If it returns 0 the
885 * dictionary can still be loaded as a content-only dictionary.
887 unsigned int ZSTD_getDictID_fromDict(const void *dict, size_t dictSize);
890 * ZSTD_getDictID_fromDDict() - returns the dictionary id stored in a ZSTD_DDict
891 * @ddict: The ddict to find the id of.
893 * Return: The dictionary id stored within `ddict` or 0 if the dictionary is not
894 * a zstd dictionary. If it returns 0 `ddict` will be loaded as a
895 * content-only dictionary.
897 unsigned int ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict);
900 * ZSTD_getDictID_fromFrame() - returns the dictionary id stored in a zstd frame
901 * @src: Source buffer. It must be a zstd encoded frame.
902 * @srcSize: The size of the source buffer. It must be at least as large as the
903 * frame header. `ZSTD_frameHeaderSize_max` is always large enough.
905 * Return: The dictionary id required to decompress the frame stored within
906 * `src` or 0 if the dictionary id could not be decoded. It can return
907 * 0 if the frame does not require a dictionary, the dictionary id
908 * wasn't stored in the frame, `src` is not a zstd frame, or `srcSize`
911 unsigned int ZSTD_getDictID_fromFrame(const void *src, size_t srcSize);
914 * struct ZSTD_frameParams - zstd frame parameters stored in the frame header
915 * @frameContentSize: The frame content size, or 0 if not present.
916 * @windowSize: The window size, or 0 if the frame is a skippable frame.
917 * @dictID: The dictionary id, or 0 if not present.
918 * @checksumFlag: Whether a checksum was used.
921 unsigned long long frameContentSize;
922 unsigned int windowSize;
924 unsigned int checksumFlag;
928 * ZSTD_getFrameParams() - extracts parameters from a zstd or skippable frame
929 * @fparamsPtr: On success the frame parameters are written here.
930 * @src: The source buffer. It must point to a zstd or skippable frame.
931 * @srcSize: The size of the source buffer. `ZSTD_frameHeaderSize_max` is
932 * always large enough to succeed.
934 * Return: 0 on success. If more data is required it returns how many bytes
935 * must be provided to make forward progress. Otherwise it returns
936 * an error, which can be checked using ZSTD_isError().
938 size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src,
941 /*-*****************************************************************************
942 * Buffer-less and synchronous inner streaming functions
944 * This is an advanced API, giving full control over buffer management, for
945 * users which need direct control over memory.
946 * But it's also a complex one, with many restrictions (documented below).
947 * Prefer using normal streaming API for an easier experience
948 ******************************************************************************/
950 /*-*****************************************************************************
951 * Buffer-less streaming compression (synchronous mode)
953 * A ZSTD_CCtx object is required to track streaming operations.
954 * Use ZSTD_initCCtx() to initialize a context.
955 * ZSTD_CCtx object can be re-used multiple times within successive compression
958 * Start by initializing a context.
959 * Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary
961 * or ZSTD_compressBegin_advanced(), for finer parameter control.
962 * It's also possible to duplicate a reference context which has already been
963 * initialized, using ZSTD_copyCCtx()
965 * Then, consume your input using ZSTD_compressContinue().
966 * There are some important considerations to keep in mind when using this
967 * advanced function :
968 * - ZSTD_compressContinue() has no internal buffer. It uses externally provided
970 * - Interface is synchronous : input is consumed entirely and produce 1+
971 * (or more) compressed blocks.
972 * - Caller must ensure there is enough space in `dst` to store compressed data
973 * under worst case scenario. Worst case evaluation is provided by
974 * ZSTD_compressBound().
975 * ZSTD_compressContinue() doesn't guarantee recover after a failed
977 * - ZSTD_compressContinue() presumes prior input ***is still accessible and
978 * unmodified*** (up to maximum distance size, see WindowLog).
979 * It remembers all previous contiguous blocks, plus one separated memory
980 * segment (which can itself consists of multiple contiguous blocks)
981 * - ZSTD_compressContinue() detects that prior input has been overwritten when
982 * `src` buffer overlaps. In which case, it will "discard" the relevant memory
983 * section from its history.
985 * Finish a frame with ZSTD_compressEnd(), which will write the last block(s)
986 * and optional checksum. It's possible to use srcSize==0, in which case, it
987 * will write a final empty block to end the frame. Without last block mark,
988 * frames will be considered unfinished (corrupted) by decoders.
990 * `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress some new
992 ******************************************************************************/
994 /*===== Buffer-less streaming compression functions =====*/
995 size_t ZSTD_compressBegin(ZSTD_CCtx *cctx, int compressionLevel);
996 size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx *cctx, const void *dict,
997 size_t dictSize, int compressionLevel);
998 size_t ZSTD_compressBegin_advanced(ZSTD_CCtx *cctx, const void *dict,
999 size_t dictSize, ZSTD_parameters params,
1000 unsigned long long pledgedSrcSize);
1001 size_t ZSTD_copyCCtx(ZSTD_CCtx *cctx, const ZSTD_CCtx *preparedCCtx,
1002 unsigned long long pledgedSrcSize);
1003 size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx *cctx, const ZSTD_CDict *cdict,
1004 unsigned long long pledgedSrcSize);
1005 size_t ZSTD_compressContinue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
1006 const void *src, size_t srcSize);
1007 size_t ZSTD_compressEnd(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
1008 const void *src, size_t srcSize);
1012 /*-*****************************************************************************
1013 * Buffer-less streaming decompression (synchronous mode)
1015 * A ZSTD_DCtx object is required to track streaming operations.
1016 * Use ZSTD_initDCtx() to initialize a context.
1017 * A ZSTD_DCtx object can be re-used multiple times.
1019 * First typical operation is to retrieve frame parameters, using
1020 * ZSTD_getFrameParams(). It fills a ZSTD_frameParams structure which provide
1021 * important information to correctly decode the frame, such as the minimum
1022 * rolling buffer size to allocate to decompress data (`windowSize`), and the
1023 * dictionary ID used.
1024 * Note: content size is optional, it may not be present. 0 means unknown.
1025 * Note that these values could be wrong, either because of data malformation,
1026 * or because an attacker is spoofing deliberate false information. As a
1027 * consequence, check that values remain within valid application range,
1028 * especially `windowSize`, before allocation. Each application can set its own
1029 * limit, depending on local restrictions. For extended interoperability, it is
1030 * recommended to support at least 8 MB.
1031 * Frame parameters are extracted from the beginning of the compressed frame.
1032 * Data fragment must be large enough to ensure successful decoding, typically
1033 * `ZSTD_frameHeaderSize_max` bytes.
1034 * Result: 0: successful decoding, the `ZSTD_frameParams` structure is filled.
1035 * >0: `srcSize` is too small, provide at least this many bytes.
1036 * errorCode, which can be tested using ZSTD_isError().
1038 * Start decompression, with ZSTD_decompressBegin() or
1039 * ZSTD_decompressBegin_usingDict(). Alternatively, you can copy a prepared
1040 * context, using ZSTD_copyDCtx().
1042 * Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue()
1044 * ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize'
1045 * to ZSTD_decompressContinue().
1046 * ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will
1049 * The result of ZSTD_decompressContinue() is the number of bytes regenerated
1050 * within 'dst' (necessarily <= dstCapacity). It can be zero, which is not an
1051 * error; it just means ZSTD_decompressContinue() has decoded some metadata
1052 * item. It can also be an error code, which can be tested with ZSTD_isError().
1054 * ZSTD_decompressContinue() needs previous data blocks during decompression, up
1055 * to `windowSize`. They should preferably be located contiguously, prior to
1056 * current block. Alternatively, a round buffer of sufficient size is also
1057 * possible. Sufficient size is determined by frame parameters.
1058 * ZSTD_decompressContinue() is very sensitive to contiguity, if 2 blocks don't
1059 * follow each other, make sure that either the compressor breaks contiguity at
1060 * the same place, or that previous contiguous segment is large enough to
1061 * properly handle maximum back-reference.
1063 * A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
1064 * Context can then be reset to start a new decompression.
1066 * Note: it's possible to know if next input to present is a header or a block,
1067 * using ZSTD_nextInputType(). This information is not required to properly
1070 * == Special case: skippable frames ==
1072 * Skippable frames allow integration of user-defined data into a flow of
1073 * concatenated frames. Skippable frames will be ignored (skipped) by a
1074 * decompressor. The format of skippable frames is as follows:
1075 * a) Skippable frame ID - 4 Bytes, Little endian format, any value from
1076 * 0x184D2A50 to 0x184D2A5F
1077 * b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
1078 * c) Frame Content - any content (User Data) of length equal to Frame Size
1079 * For skippable frames ZSTD_decompressContinue() always returns 0.
1080 * For skippable frames ZSTD_getFrameParams() returns fparamsPtr->windowLog==0
1081 * what means that a frame is skippable.
1082 * Note: If fparamsPtr->frameContentSize==0, it is ambiguous: the frame might
1083 * actually be a zstd encoded frame with no content. For purposes of
1084 * decompression, it is valid in both cases to skip the frame using
1085 * ZSTD_findFrameCompressedSize() to find its size in bytes.
1086 * It also returns frame size as fparamsPtr->frameContentSize.
1087 ******************************************************************************/
1089 /*===== Buffer-less streaming decompression functions =====*/
1090 size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx);
1091 size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict,
1093 void ZSTD_copyDCtx(ZSTD_DCtx *dctx, const ZSTD_DCtx *preparedDCtx);
1094 size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx);
1095 size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
1096 const void *src, size_t srcSize);
1098 ZSTDnit_frameHeader,
1099 ZSTDnit_blockHeader,
1103 ZSTDnit_skippableFrame
1104 } ZSTD_nextInputType_e;
1105 ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx);
1107 /*-*****************************************************************************
1110 * Block functions produce and decode raw zstd blocks, without frame metadata.
1111 * Frame metadata cost is typically ~18 bytes, which can be non-negligible for
1112 * very small blocks (< 100 bytes). User will have to take in charge required
1113 * information to regenerate data, such as compressed and content sizes.
1115 * A few rules to respect:
1116 * - Compressing and decompressing require a context structure
1117 * + Use ZSTD_initCCtx() and ZSTD_initDCtx()
1118 * - It is necessary to init context before starting
1119 * + compression : ZSTD_compressBegin()
1120 * + decompression : ZSTD_decompressBegin()
1121 * + variants _usingDict() are also allowed
1122 * + copyCCtx() and copyDCtx() work too
1123 * - Block size is limited, it must be <= ZSTD_getBlockSizeMax()
1124 * + If you need to compress more, cut data into multiple blocks
1125 * + Consider using the regular ZSTD_compress() instead, as frame metadata
1126 * costs become negligible when source size is large.
1127 * - When a block is considered not compressible enough, ZSTD_compressBlock()
1128 * result will be zero. In which case, nothing is produced into `dst`.
1129 * + User must test for such outcome and deal directly with uncompressed data
1130 * + ZSTD_decompressBlock() doesn't accept uncompressed data as input!!!
1131 * + In case of multiple successive blocks, decoder must be informed of
1132 * uncompressed block existence to follow proper history. Use
1133 * ZSTD_insertBlock() in such a case.
1134 ******************************************************************************/
1136 /* Define for static allocation */
1137 #define ZSTD_BLOCKSIZE_ABSOLUTEMAX (128 * 1024)
1138 /*===== Raw zstd block functions =====*/
1139 size_t ZSTD_getBlockSizeMax(ZSTD_CCtx *cctx);
1140 size_t ZSTD_compressBlock(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity,
1141 const void *src, size_t srcSize);
1142 size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity,
1143 const void *src, size_t srcSize);
1144 size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart,