Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / zstd / 33eb7ac6b6e152ce9263fdb1f84eb9bf50abe1eb / . / doc / zstd_manual.html
blob: 9b0973164f40b90550d6ba5e123cb324fdda113a [file] [log] [blame]
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]1<html>
2<head>
3<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
Yann Colletfa8dadb2017-05-08 18:24:16 -0700[diff] [blame]4<title>zstd 1.3.0 Manual</title>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]5</head>
6<body>
Yann Colletfa8dadb2017-05-08 18:24:16 -0700[diff] [blame]7<h1>zstd 1.3.0 Manual</h1>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]8<hr>
9<a name="Contents"></a><h2>Contents</h2>
10<ol>
11<li><a href="#Chapter1">Introduction</a></li>
12<li><a href="#Chapter2">Version</a></li>
13<li><a href="#Chapter3">Simple API</a></li>
14<li><a href="#Chapter4">Explicit memory management</a></li>
15<li><a href="#Chapter5">Simple dictionary API</a></li>
16<li><a href="#Chapter6">Fast dictionary API</a></li>
17<li><a href="#Chapter7">Streaming</a></li>
18<li><a href="#Chapter8">Streaming compression - HowTo</a></li>
19<li><a href="#Chapter9">Streaming decompression - HowTo</a></li>
20<li><a href="#Chapter10">START OF ADVANCED AND EXPERIMENTAL FUNCTIONS</a></li>
21<li><a href="#Chapter11">Advanced types</a></li>
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]22<li><a href="#Chapter12">Frame size functions</a></li>
23<li><a href="#Chapter13">Context memory usage</a></li>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]24<li><a href="#Chapter14">Advanced compression functions</a></li>
25<li><a href="#Chapter15">Advanced decompression functions</a></li>
26<li><a href="#Chapter16">Advanced streaming functions</a></li>
27<li><a href="#Chapter17">Buffer-less and synchronous inner streaming functions</a></li>
28<li><a href="#Chapter18">Buffer-less streaming compression (synchronous mode)</a></li>
29<li><a href="#Chapter19">Buffer-less streaming decompression (synchronous mode)</a></li>
30<li><a href="#Chapter20">Block functions</a></li>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]31</ol>
32<hr>
33<a name="Chapter1"></a><h2>Introduction</h2><pre>
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]34 zstd, short for Zstandard, is a fast lossless compression algorithm, targeting real-time compression scenarios
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]35 at zlib-level and better compression ratios. The zstd compression library provides in-memory compression and
36 decompression functions. The library supports compression levels from 1 up to ZSTD_maxCLevel() which is 22.
Yann Collet831b4892017-02-23 23:09:10 -0800[diff] [blame]37 Levels >= 20, labeled `--ultra`, should be used with caution, as they require more memory.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]38 Compression can be done in:
39 - a single step (described as Simple API)
40 - a single step, reusing a context (described as Explicit memory management)
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]41 - unbounded multiple steps (described as Streaming compression)
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]42 The compression ratio achievable on small data can be highly improved using compression with a dictionary in:
43 - a single step (described as Simple dictionary API)
44 - a single step, reusing a dictionary (described as Fast dictionary API)
45
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]46 Advanced experimental functions can be accessed using #define ZSTD_STATIC_LINKING_ONLY before including zstd.h.
47 These APIs shall never be used with a dynamic library.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]48 They are not "stable", their definition may change in the future. Only static linking is allowed.
49<BR></pre>
50
51<a name="Chapter2"></a><h2>Version</h2><pre></pre>
52
Przemyslaw Skibinski4da53212016-12-07 11:18:40 +0100[diff] [blame]53<pre><b>unsigned ZSTD_versionNumber(void); </b>/**< library version number; to be used when checking dll version */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]54</b></pre><BR>
55<a name="Chapter3"></a><h2>Simple API</h2><pre></pre>
56
57<pre><b>size_t ZSTD_compress( void* dst, size_t dstCapacity,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]58 const void* src, size_t srcSize,
59 int compressionLevel);
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]60</b><p> Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
61 Hint : compression runs faster if `dstCapacity` >= `ZSTD_compressBound(srcSize)`.
62 @return : compressed size written into `dst` (<= `dstCapacity),
63 or an error code if it fails (which can be tested using ZSTD_isError()).
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]64</p></pre><BR>
65
66<pre><b>size_t ZSTD_decompress( void* dst, size_t dstCapacity,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]67 const void* src, size_t compressedSize);
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]68</b><p> `compressedSize` : must be the _exact_ size of some number of compressed and/or skippable frames.
69 `dstCapacity` is an upper bound of originalSize.
70 If user cannot imply a maximum upper bound, it's better to use streaming mode to decompress data.
71 @return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
72 or an errorCode if it fails (which can be tested using ZSTD_isError()).
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]73</p></pre><BR>
74
75<pre><b>unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]76</b><p> NOTE: This function is planned to be obsolete, in favour of ZSTD_getFrameContentSize.
77 ZSTD_getFrameContentSize functions the same way, returning the decompressed size of a single
78 frame, but distinguishes empty frames from frames with an unknown size, or errors.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]79
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]80 Additionally, ZSTD_findDecompressedSize can be used instead. It can handle multiple
81 concatenated frames in one buffer, and so is more general.
82 As a result however, it requires more computation and entire frames to be passed to it,
83 as opposed to ZSTD_getFrameContentSize which requires only a single frame's header.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]84
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]85 'src' is the start of a zstd compressed frame.
86 @return : content size to be decompressed, as a 64-bits value _if known_, 0 otherwise.
87 note 1 : decompressed size is an optional field, that may not be present, especially in streaming mode.
88 When `return==0`, data to decompress could be any size.
89 In which case, it's necessary to use streaming mode to decompress data.
90 Optionally, application can still use ZSTD_decompress() while relying on implied limits.
91 (For example, data may be necessarily cut into blocks <= 16 KB).
92 note 2 : decompressed size is always present when compression is done with ZSTD_compress()
93 note 3 : decompressed size can be very large (64-bits value),
94 potentially larger than what local system can handle as a single memory segment.
95 In which case, it's necessary to use streaming mode to decompress data.
96 note 4 : If source is untrusted, decompressed size could be wrong or intentionally modified.
97 Always ensure result fits within application's authorized limits.
98 Each application can set its own limits.
99 note 5 : when `return==0`, if precise failure cause is needed, use ZSTD_getFrameParams() to know more.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]100</p></pre><BR>
101
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]102<h3>Helper functions</h3><pre></pre><b><pre>int ZSTD_maxCLevel(void); </b>/*!< maximum compression level available */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]103size_t ZSTD_compressBound(size_t srcSize); </b>/*!< maximum compressed size in worst case scenario */<b>
104unsigned ZSTD_isError(size_t code); </b>/*!< tells if a `size_t` function result is an error code */<b>
105const char* ZSTD_getErrorName(size_t code); </b>/*!< provides readable string from an error code */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]106</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]107<a name="Chapter4"></a><h2>Explicit memory management</h2><pre></pre>
108
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]109<h3>Compression context</h3><pre> When compressing many times,
110 it is recommended to allocate a context just once, and re-use it for each successive compression operation.
111 This will make workload friendlier for system's memory.
112 Use one context per thread for parallel execution in multi-threaded environments.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]113</pre><b><pre>typedef struct ZSTD_CCtx_s ZSTD_CCtx;
114ZSTD_CCtx* ZSTD_createCCtx(void);
115size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx);
116</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]117<pre><b>size_t ZSTD_compressCCtx(ZSTD_CCtx* ctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize, int compressionLevel);
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]118</b><p> Same as ZSTD_compress(), requires an allocated ZSTD_CCtx (see ZSTD_createCCtx()).
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]119</p></pre><BR>
120
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]121<h3>Decompression context</h3><pre> When decompressing many times,
122 it is recommended to allocate a context just once, and re-use it for each successive compression operation.
123 This will make workload friendlier for system's memory.
124 Use one context per thread for parallel execution in multi-threaded environments.
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]125</pre><b><pre>typedef struct ZSTD_DCtx_s ZSTD_DCtx;
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]126ZSTD_DCtx* ZSTD_createDCtx(void);
127size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx);
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]128</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]129<pre><b>size_t ZSTD_decompressDCtx(ZSTD_DCtx* ctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
Yann Collete42afbc2017-04-26 11:39:35 -0700[diff] [blame]130</b><p> Same as ZSTD_decompress(), requires an allocated ZSTD_DCtx (see ZSTD_createDCtx()).
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]131</p></pre><BR>
132
133<a name="Chapter5"></a><h2>Simple dictionary API</h2><pre></pre>
134
135<pre><b>size_t ZSTD_compress_usingDict(ZSTD_CCtx* ctx,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]136 void* dst, size_t dstCapacity,
137 const void* src, size_t srcSize,
138 const void* dict,size_t dictSize,
139 int compressionLevel);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]140</b><p> Compression using a predefined Dictionary (see dictBuilder/zdict.h).
Nick Terrelld82efd82016-11-02 16:47:53 -0700[diff] [blame]141 Note : This function loads the dictionary, resulting in significant startup delay.
142 Note : When `dict == NULL || dictSize < 8` no dictionary is used.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]143</p></pre><BR>
144
145<pre><b>size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]146 void* dst, size_t dstCapacity,
147 const void* src, size_t srcSize,
148 const void* dict,size_t dictSize);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]149</b><p> Decompression using a predefined Dictionary (see dictBuilder/zdict.h).
150 Dictionary must be identical to the one used during compression.
Nick Terrelld82efd82016-11-02 16:47:53 -0700[diff] [blame]151 Note : This function loads the dictionary, resulting in significant startup delay.
152 Note : When `dict == NULL || dictSize < 8` no dictionary is used.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]153</p></pre><BR>
154
155<a name="Chapter6"></a><h2>Fast dictionary API</h2><pre></pre>
156
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]157<pre><b>ZSTD_CDict* ZSTD_createCDict(const void* dictBuffer, size_t dictSize, int compressionLevel);
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]158</b><p> When compressing multiple messages / blocks with the same dictionary, it's recommended to load it just once.
159 ZSTD_createCDict() will create a digested dictionary, ready to start future compression operations without startup delay.
160 ZSTD_CDict can be created once and used by multiple threads concurrently, as its usage is read-only.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]161 `dictBuffer` can be released after ZSTD_CDict creation, as its content is copied within CDict
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]162</p></pre><BR>
163
164<pre><b>size_t ZSTD_freeCDict(ZSTD_CDict* CDict);
Nick Terrelld82efd82016-11-02 16:47:53 -0700[diff] [blame]165</b><p> Function frees memory allocated by ZSTD_createCDict().
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]166</p></pre><BR>
167
168<pre><b>size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]169 void* dst, size_t dstCapacity,
170 const void* src, size_t srcSize,
171 const ZSTD_CDict* cdict);
Yann Collet715b9aa2017-04-18 13:55:53 -0700[diff] [blame]172</b><p> Compression using a digested Dictionary.
173 Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times.
174 Note that compression level is decided during dictionary creation.
175 Frame parameters are hardcoded (dictID=yes, contentSize=yes, checksum=no)
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]176</p></pre><BR>
177
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]178<pre><b>ZSTD_DDict* ZSTD_createDDict(const void* dictBuffer, size_t dictSize);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]179</b><p> Create a digested dictionary, ready to start decompression operation without startup delay.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]180 dictBuffer can be released after DDict creation, as its content is copied inside DDict
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]181</p></pre><BR>
182
183<pre><b>size_t ZSTD_freeDDict(ZSTD_DDict* ddict);
184</b><p> Function frees memory allocated with ZSTD_createDDict()
185</p></pre><BR>
186
187<pre><b>size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]188 void* dst, size_t dstCapacity,
189 const void* src, size_t srcSize,
190 const ZSTD_DDict* ddict);
Nick Terrelld82efd82016-11-02 16:47:53 -0700[diff] [blame]191</b><p> Decompression using a digested Dictionary.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]192 Faster startup than ZSTD_decompress_usingDict(), recommended when same dictionary is used multiple times.
193</p></pre><BR>
194
195<a name="Chapter7"></a><h2>Streaming</h2><pre></pre>
196
197<pre><b>typedef struct ZSTD_inBuffer_s {
198 const void* src; </b>/**< start of input buffer */<b>
199 size_t size; </b>/**< size of input buffer */<b>
200 size_t pos; </b>/**< position where reading stopped. Will be updated. Necessarily 0 <= pos <= size */<b>
201} ZSTD_inBuffer;
202</b></pre><BR>
203<pre><b>typedef struct ZSTD_outBuffer_s {
204 void* dst; </b>/**< start of output buffer */<b>
205 size_t size; </b>/**< size of output buffer */<b>
206 size_t pos; </b>/**< position where writing stopped. Will be updated. Necessarily 0 <= pos <= size */<b>
207} ZSTD_outBuffer;
208</b></pre><BR>
209<a name="Chapter8"></a><h2>Streaming compression - HowTo</h2><pre>
210 A ZSTD_CStream object is required to track streaming operation.
211 Use ZSTD_createCStream() and ZSTD_freeCStream() to create/release resources.
212 ZSTD_CStream objects can be reused multiple times on consecutive compression operations.
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]213 It is recommended to re-use ZSTD_CStream in situations where many streaming operations will be achieved consecutively,
214 since it will play nicer with system's memory, by re-using already allocated memory.
215 Use one separate ZSTD_CStream per thread for parallel execution.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]216
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]217 Start a new compression by initializing ZSTD_CStream.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]218 Use ZSTD_initCStream() to start a new compression operation.
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]219 Use ZSTD_initCStream_usingDict() or ZSTD_initCStream_usingCDict() for a compression which requires a dictionary (experimental section)
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]220
221 Use ZSTD_compressStream() repetitively to consume input stream.
222 The function will automatically update both `pos` fields.
223 Note that it may not consume the entire input, in which case `pos < size`,
224 and it's up to the caller to present again remaining data.
225 @return : a size hint, preferred nb of bytes to use as input for next function call
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]226 or an error code, which can be tested using ZSTD_isError().
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]227 Note 1 : it's just a hint, to help latency a little, any other value will work fine.
228 Note 2 : size hint is guaranteed to be <= ZSTD_CStreamInSize()
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]229
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]230 At any moment, it's possible to flush whatever data remains within internal buffer, using ZSTD_flushStream().
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]231 `output->pos` will be updated.
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]232 Note that some content might still be left within internal buffer if `output->size` is too small.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]233 @return : nb of bytes still present within internal buffer (0 if it's empty)
234 or an error code, which can be tested using ZSTD_isError().
235
236 ZSTD_endStream() instructs to finish a frame.
237 It will perform a flush and write frame epilogue.
238 The epilogue is required for decoders to consider a frame completed.
239 Similar to ZSTD_flushStream(), it may not be able to flush the full content if `output->size` is too small.
240 In which case, call again ZSTD_endStream() to complete the flush.
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]241 @return : nb of bytes still present within internal buffer (0 if it's empty, hence compression completed)
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]242 or an error code, which can be tested using ZSTD_isError().
243
244
245<BR></pre>
246
Yann Collet0be6fd32017-05-08 16:08:01 -0700[diff] [blame]247<pre><b>typedef ZSTD_CCtx ZSTD_CStream; </b>/**< CCtx and CStream are effectively same object */<b>
248</b></pre><BR>
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]249<h3>ZSTD_CStream management functions</h3><pre></pre><b><pre>ZSTD_CStream* ZSTD_createCStream(void);
250size_t ZSTD_freeCStream(ZSTD_CStream* zcs);
251</pre></b><BR>
252<h3>Streaming compression functions</h3><pre></pre><b><pre>size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel);
253size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
254size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
255size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
256</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]257<pre><b>size_t ZSTD_CStreamInSize(void); </b>/**< recommended size for input buffer */<b>
258</b></pre><BR>
259<pre><b>size_t ZSTD_CStreamOutSize(void); </b>/**< recommended size for output buffer. Guarantee to successfully flush at least one complete compressed block in all circumstances. */<b>
260</b></pre><BR>
261<a name="Chapter9"></a><h2>Streaming decompression - HowTo</h2><pre>
262 A ZSTD_DStream object is required to track streaming operations.
263 Use ZSTD_createDStream() and ZSTD_freeDStream() to create/release resources.
264 ZSTD_DStream objects can be re-used multiple times.
265
266 Use ZSTD_initDStream() to start a new decompression operation,
267 or ZSTD_initDStream_usingDict() if decompression requires a dictionary.
268 @return : recommended first input size
269
270 Use ZSTD_decompressStream() repetitively to consume your input.
271 The function will update both `pos` fields.
272 If `input.pos < input.size`, some input has not been consumed.
273 It's up to the caller to present again remaining data.
274 If `output.pos < output.size`, decoder has flushed everything it could.
275 @return : 0 when a frame is completely decoded and fully flushed,
276 an error code, which can be tested using ZSTD_isError(),
Przemyslaw Skibinski4da53212016-12-07 11:18:40 +0100[diff] [blame]277 any other value > 0, which means there is still some decoding to do to complete current frame.
278 The return value is a suggested next input size (a hint to improve latency) that will never load more than the current frame.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]279
280<BR></pre>
281
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]282<h3>ZSTD_DStream management functions</h3><pre></pre><b><pre>ZSTD_DStream* ZSTD_createDStream(void);
283size_t ZSTD_freeDStream(ZSTD_DStream* zds);
284</pre></b><BR>
285<h3>Streaming decompression functions</h3><pre></pre><b><pre>size_t ZSTD_initDStream(ZSTD_DStream* zds);
286size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
287</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]288<pre><b>size_t ZSTD_DStreamInSize(void); </b>/*!< recommended size for input buffer */<b>
289</b></pre><BR>
290<pre><b>size_t ZSTD_DStreamOutSize(void); </b>/*!< recommended size for output buffer. Guarantee to successfully flush at least one complete block in all circumstances. */<b>
291</b></pre><BR>
292<a name="Chapter10"></a><h2>START OF ADVANCED AND EXPERIMENTAL FUNCTIONS</h2><pre> The definitions in this section are considered experimental.
293 They should never be used with a dynamic library, as they may change in the future.
294 They are provided for advanced usages.
295 Use them only in association with static linking.
296
297<BR></pre>
298
299<a name="Chapter11"></a><h2>Advanced types</h2><pre></pre>
300
Nick Terrelleeb31ee2017-03-09 11:44:25 -0800[diff] [blame]301<pre><b>typedef enum { ZSTD_fast, ZSTD_dfast, ZSTD_greedy, ZSTD_lazy, ZSTD_lazy2, ZSTD_btlazy2, ZSTD_btopt, ZSTD_btultra } ZSTD_strategy; </b>/* from faster to stronger */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]302</b></pre><BR>
303<pre><b>typedef struct {
304 unsigned windowLog; </b>/**< largest match distance : larger == more compression, more memory needed during decompression */<b>
305 unsigned chainLog; </b>/**< fully searched segment : larger == more compression, slower, more memory (useless for fast) */<b>
306 unsigned hashLog; </b>/**< dispatch table : larger == faster, more memory */<b>
307 unsigned searchLog; </b>/**< nb of searches : larger == more compression, slower */<b>
308 unsigned searchLength; </b>/**< match length searched : larger == faster decompression, sometimes less compression */<b>
309 unsigned targetLength; </b>/**< acceptable match size for optimal parser (only) : larger == more compression, slower */<b>
310 ZSTD_strategy strategy;
311} ZSTD_compressionParameters;
312</b></pre><BR>
313<pre><b>typedef struct {
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]314 unsigned contentSizeFlag; </b>/**< 1: content size will be in frame header (when known) */<b>
315 unsigned checksumFlag; </b>/**< 1: generate a 32-bits checksum at end of frame, for error detection */<b>
316 unsigned noDictIDFlag; </b>/**< 1: no dictID will be saved into frame header (if dictionary compression) */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]317} ZSTD_frameParameters;
318</b></pre><BR>
319<pre><b>typedef struct {
320 ZSTD_compressionParameters cParams;
321 ZSTD_frameParameters fParams;
322} ZSTD_parameters;
323</b></pre><BR>
Yann Colletf16f4492017-05-09 16:18:17 -0700[diff] [blame]324<pre><b>typedef struct {
325 unsigned long long frameContentSize;
326 unsigned windowSize;
327 unsigned dictID;
328 unsigned checksumFlag;
329} ZSTD_frameHeader;
330</b></pre><BR>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]331<h3>Custom memory allocation functions</h3><pre></pre><b><pre>typedef void* (*ZSTD_allocFunction) (void* opaque, size_t size);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]332typedef void (*ZSTD_freeFunction) (void* opaque, void* address);
333typedef struct { ZSTD_allocFunction customAlloc; ZSTD_freeFunction customFree; void* opaque; } ZSTD_customMem;
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]334</pre></b><BR>
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]335<a name="Chapter12"></a><h2>Frame size functions</h2><pre></pre>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]336
Yann Collet831b4892017-02-23 23:09:10 -0800[diff] [blame]337<pre><b>size_t ZSTD_findFrameCompressedSize(const void* src, size_t srcSize);
338</b><p> `src` should point to the start of a ZSTD encoded frame or skippable frame
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]339 `srcSize` must be at least as large as the frame
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]340 @return : the compressed size of the frame pointed to by `src`,
341 suitable to pass to `ZSTD_decompress` or similar,
342 or an error code if given invalid input.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]343</p></pre><BR>
344
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]345<pre><b>#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
346#define ZSTD_CONTENTSIZE_ERROR (0ULL - 2)
347unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);
348</b><p> `src` should point to the start of a ZSTD encoded frame.
349 `srcSize` must be at least as large as the frame header.
350 A value >= `ZSTD_frameHeaderSize_max` is guaranteed to be large enough.
351 @return : - decompressed size of the frame pointed to be `src` if known
352 - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
353 - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small)
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]354</p></pre><BR>
355
356<pre><b>unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]357</b><p> `src` should point the start of a series of ZSTD encoded and/or skippable frames
358 `srcSize` must be the _exact_ size of this series
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]359 (i.e. there should be a frame boundary exactly `srcSize` bytes after `src`)
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]360 @return : - decompressed size of all data in all successive frames
361 - if the decompressed size cannot be determined: ZSTD_CONTENTSIZE_UNKNOWN
362 - if an error occurred: ZSTD_CONTENTSIZE_ERROR
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]363
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]364 note 1 : decompressed size is an optional field, that may not be present, especially in streaming mode.
365 When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
366 In which case, it's necessary to use streaming mode to decompress data.
367 Optionally, application can still use ZSTD_decompress() while relying on implied limits.
368 (For example, data may be necessarily cut into blocks <= 16 KB).
369 note 2 : decompressed size is always present when compression is done with ZSTD_compress()
370 note 3 : decompressed size can be very large (64-bits value),
371 potentially larger than what local system can handle as a single memory segment.
372 In which case, it's necessary to use streaming mode to decompress data.
373 note 4 : If source is untrusted, decompressed size could be wrong or intentionally modified.
374 Always ensure result fits within application's authorized limits.
375 Each application can set its own limits.
376 note 5 : ZSTD_findDecompressedSize handles multiple frames, and so it must traverse the input to
377 read each contained frame header. This is efficient as most of the data is skipped,
378 however it does mean that all frame data must be present and valid.
379</p></pre><BR>
380
381<a name="Chapter13"></a><h2>Context memory usage</h2><pre></pre>
382
383<pre><b>size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx);
384size_t ZSTD_sizeof_DCtx(const ZSTD_DCtx* dctx);
385size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs);
386size_t ZSTD_sizeof_DStream(const ZSTD_DStream* zds);
387size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict);
388size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);
389</b><p> These functions give the current memory usage of selected object.
390 Object memory usage can evolve if it's re-used multiple times.
391</p></pre><BR>
392
393<pre><b>size_t ZSTD_estimateCCtxSize(ZSTD_compressionParameters cParams);
394size_t ZSTD_estimateDCtxSize(void);
395</b><p> These functions make it possible to estimate memory usage
396 of a future target object, before its allocation,
397 given a set of parameters, which vary depending on target object.
398 The objective is to guide decision before allocation.
399</p></pre><BR>
400
401<pre><b>size_t ZSTD_estimateCStreamSize(ZSTD_compressionParameters cParams);
Yann Colletf16f4492017-05-09 16:18:17 -0700[diff] [blame]402size_t ZSTD_estimateDStreamSize(ZSTD_frameHeader fHeader);
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]403</b><p> Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
404 an internal ?Dict will be created, which size is not estimated.
405 In this case, get additional size by using ZSTD_estimate?DictSize
406</p></pre><BR>
407
408<pre><b>size_t ZSTD_estimateCDictSize(ZSTD_compressionParameters cParams, size_t dictSize);
409size_t ZSTD_estimateDDictSize(size_t dictSize);
410</b><p> Note : if dictionary is created "byReference", reduce estimation by dictSize
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]411</p></pre><BR>
412
413<a name="Chapter14"></a><h2>Advanced compression functions</h2><pre></pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]414
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]415<pre><b>ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem);
416</b><p> Create a ZSTD compression context using external alloc and free functions
417</p></pre><BR>
418
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]419<pre><b>typedef enum {
Yann Collet14312d82017-02-23 23:42:12 -0800[diff] [blame]420 ZSTD_p_forceWindow, </b>/* Force back-references to remain < windowSize, even when referencing Dictionary content (default:0) */<b>
421 ZSTD_p_forceRawDict </b>/* Force loading dictionary in "content-only" mode (no header analysis) */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]422} ZSTD_CCtxParameter;
423</b></pre><BR>
424<pre><b>size_t ZSTD_setCCtxParameter(ZSTD_CCtx* cctx, ZSTD_CCtxParameter param, unsigned value);
425</b><p> Set advanced parameters, selected through enum ZSTD_CCtxParameter
426 @result : 0, or an error code (which can be tested with ZSTD_isError())
427</p></pre><BR>
428
429<pre><b>ZSTD_CDict* ZSTD_createCDict_byReference(const void* dictBuffer, size_t dictSize, int compressionLevel);
430</b><p> Create a digested dictionary for compression
431 Dictionary content is simply referenced, and therefore stays in dictBuffer.
432 It is important that dictBuffer outlives CDict, it must remain read accessible throughout the lifetime of CDict
433</p></pre><BR>
434
435<pre><b>ZSTD_CDict* ZSTD_createCDict_advanced(const void* dict, size_t dictSize, unsigned byReference,
Yann Collet31533ba2017-04-27 00:29:04 -0700[diff] [blame]436 ZSTD_compressionParameters cParams, ZSTD_customMem customMem);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]437</b><p> Create a ZSTD_CDict using external alloc and free, and customized compression parameters
438</p></pre><BR>
439
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]440<pre><b>ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
441</b><p> @return ZSTD_compressionParameters structure for a selected compression level and estimated srcSize.
442 `estimatedSrcSize` value is optional, select 0 if not known
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]443</p></pre><BR>
444
Yann Colletdc993122016-12-14 14:53:47 +0100[diff] [blame]445<pre><b>ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);
446</b><p> same as ZSTD_getCParams(), but @return a full `ZSTD_parameters` object instead of sub-component `ZSTD_compressionParameters`.
447 All fields of `ZSTD_frameParameters` are set to default (0)
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]448</p></pre><BR>
449
450<pre><b>size_t ZSTD_checkCParams(ZSTD_compressionParameters params);
451</b><p> Ensure param values remain within authorized range
452</p></pre><BR>
453
454<pre><b>ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize);
455</b><p> optimize params for a given `srcSize` and `dictSize`.
456 both values are optional, select `0` if unknown.
457</p></pre><BR>
458
Yann Colletf4bd8572017-04-27 11:31:55 -0700[diff] [blame]459<pre><b>size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx,
460 void* dst, size_t dstCapacity,
461 const void* src, size_t srcSize,
462 const void* dict,size_t dictSize,
463 ZSTD_parameters params);
464</b><p> Same as ZSTD_compress_usingDict(), with fine-tune control over each compression parameter
465</p></pre><BR>
466
467<pre><b>size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
468 void* dst, size_t dstCapacity,
469 const void* src, size_t srcSize,
470 const ZSTD_CDict* cdict, ZSTD_frameParameters fParams);
Yann Collet77bf59e2017-04-27 11:43:04 -0700[diff] [blame]471</b><p> Same as ZSTD_compress_usingCDict(), with fine-tune control over frame parameters
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]472</p></pre><BR>
473
Yann Collet33eb7ac2017-05-12 12:36:11 -0700[diff] [blame^]474<h3>New experimental advanced parameters API</h3><pre></pre><b><pre></pre></b><BR>
475<pre><b>typedef enum {
476 </b>/* compression parameters */<b>
477 ZSTD_p_compressionLevel=100, </b>/* Update all compression parameters according to pre-defined cLevel table (default:3) */<b>
478 ZSTD_p_windowLog, </b>/* Maximum allowed back-reference distance, expressed as power of 2.<b>
479 * Must be clamped between ZSTD_WINDOWLOG_MIN and ZSTD_WINDOWLOG_MAX.
480 * default value : set through compressionLevel */
481 ZSTD_p_hashLog, </b>/* Size of the probe table, as a power of 2.<b>
482 * Resulting table size is (1 << (hashLog+2)).
483 * Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX.
484 * Larger tables improve compression ratio of strategies <= dFast,
485 * and improve speed of strategies > dFast */
486 ZSTD_p_chainLog, </b>/* Size of the full-search table, as a power of 2.<b>
487 * Resulting table size is (1 << (chainLog+2)).
488 * Larger tables result in better and slower compression.
489 * This parameter is useless when using "fast" strategy */
490 ZSTD_p_searchLog, </b>/* Number of search attempts, as a power of 2.<b>
491 * More attempts result in better and slower compression.
492 * This parameter is useless when using "fast" and "dFast" strategies */
493 ZSTD_p_minMatchLength, </b>/* Minimum match size (except for repeat-matches, which limit is hard-coded).<b>
494 * Larger values make compression and decompression faster, but decrease compression ratio
495 * Must be clamped between ZSTD_SEARCHLENGTH_MIN and ZSTD_SEARCHLENGTH_MAX.
496 * Note that currently, for all strategies < btopt, effective minimum is 4.
497 * Note that currently, for all strategies > fast, effective maximum is 6. */
498 ZSTD_p_targetLength, </b>/* Only useful for strategies >= btopt.<b>
499 * Length of Match considered "good enough" to stop search.
500 * Larger values make compression stronger and slower. */
501 ZSTD_p_compressionStrategy, </b>/* See ZSTD_strategy enum definition.<b>
502 * Cast selected strategy into unsigned for ZSTD_CCtx_setParameter() compatibility.
503 * The higher the value of selected strategy, the more complex it is,
504 * resulting in stronger and slower compression */
505#if 0
506 ZSTD_p_windowSize, </b>/* Maximum allowed back-reference distance.<b>
507 * Can be set to a more precise value than windowLog.
508 * Will be transparently reduced to closest possible inferior value
509 * (see Zstandard compression format) */
510 </b>/* Not ready yet ! */<b>
511#endif
512
513 </b>/* frame parameters */<b>
514 ZSTD_p_contentSizeFlag=200, </b>/* Content size is written into frame header _whenever known_ (default:1) */<b>
515 ZSTD_p_contentChecksumFlag, </b>/* A 32-bits content checksum is calculated and written at end of frame (default:0) */<b>
516 ZSTD_p_dictIDFlag, </b>/* When applicable, dictID of dictionary is provided in frame header (default:1) */<b>
517
518 </b>/* dictionary parameters */<b>
519 ZSTD_p_refDictContent=300, </b>/* Content of dictionary content will be referenced, instead of copied (default:0).<b>
520 * This avoids duplicating dictionary content.
521 * But it also requires that dictionary buffer outlives its user (CCtx or CDict) */
522 </b>/* Not ready yet ! */<b>
523 ZSTD_p_rawContentDict, </b>/* load dictionary in "content-only" mode (no header analysis) (default:0) */<b>
524 </b>/* question : should there be an option to load dictionary only in zstd format, rejecting others with an error code ? */<b>
525
526#if 0
527 </b>/* multi-threading parameters (not ready yet !) */<b>
528 ZSTD_p_nbThreads=400, </b>/* Select how many threads a compression job can spawn (default:1)<b>
529 * More threads improve speed, but increases also memory usage */
530 ZSTDMT_p_jobSize, </b>/* Size of a compression job. Each job is compressed in parallel.<b>
531 * 0 means default, which is dynamically determined based on compression parameters.
532 * Job size must be a minimum of overlapSize, or 1 KB, whichever is largest
533 * The minimum size is automatically and transparently enforced */
534 ZSTDMT_p_overlapSizeLog, </b>/* Size of previous input reloaded at the beginning of each job.<b>
535 * 0 => no overlap, 6(default) => use 1/8th of windowSize, >=9 => use full windowSize */
536#endif
537
538 </b>/* advanced parameters - may not remain available after API update */<b>
539 ZSTD_p_forceMaxWindow=1100, </b>/* Force back-references to remain < windowSize,<b>
540 * even when referencing into Dictionary content
541 * default : 0 when using a CDict, 1 when using a Prefix */
542} ZSTD_cParameter;
543</b></pre><BR>
544<pre><b>size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, unsigned value);
545</b><p> Set one compression parameter, selected by enum ZSTD_cParameter.
546 @result : 0, or an error code (which can be tested with ZSTD_isError())
547</p></pre><BR>
548
549<pre><b>size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize);
550</b><p> Total input data size to be compressed into a single frame.
551 This value will be controlled at the end, and result in error if not respected.
552 @result : 0, or an error code (which can be tested with ZSTD_isError()).
553 Note 1 : 0 means zero, empty.
554 In order to mean "unknown content size", pass constant ZSTD_CONTENTSIZE_UNKNOWN.
555 Note that ZSTD_CONTENTSIZE_UNKNOWN is default value for all new compression jobs.
556 Note 2 : If all data is provided and consumed in a single round,
557 this value is overriden by srcSize instead.
558</p></pre><BR>
559
560<pre><b>size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize); </b>/* Not ready yet ! */<b>
561</b><p> Reference a prefix (content-only dictionary) to bootstrap next compression job.
562 Decompression will have to use same prefix.
563 @result : 0, or an error code (which can be tested with ZSTD_isError()).
564 Special : Adding a NULL (or 0-size) dictionary invalidates any previous prefix, meaning "return to no-dictionary mode".
565 Note 1 : Prefix content is referenced. It must outlive compression job.
566 Note 2 : Referencing a prefix involves building tables, which are dependent on compression parameters.
567 For this reason, compression parameters cannot be changed anymore after loading a prefix.
568 It's also a CPU-heavy operation, with non-negligible impact on latency.
569 Note 3 : Prefix is only used once. Tables are discarded at end of compression job.
570 If there is a need to use same prefix multiple times, consider embedding it into a ZSTD_CDict
571</p></pre><BR>
572
573<pre><b>ZSTD_CDict* ZSTD_CDict_createEmpty(void); </b>/* Not ready yet ! */<b>
574size_t ZSTD_CDict_setParameter(ZSTD_CDict* cdict, ZSTD_cParameter param, unsigned value); </b>/* Not ready yet ! */<b>
575size_t ZSTD_CDict_loadDictionary(ZSTD_CDict* cdict, const void* dict, size_t dictSize); </b>/* Not ready yet ! */<b>
576</b><p> Create a CDict object which is still mutable after creation.
577 It allows usage of ZSTD_CDict_setParameter().
578 Once all compression parameters are selected,
579 it's possible to load the target dictionary, using ZSTD_CDict_loadDictionary().
580 Dictionary content will be copied internally, except if ZSTD_p_refDictContent is used.
581 After loading the dictionary, no more change is possible.
582 The only remaining operation is to free CDict object.
583 Note : An unfinished CDict behaves the same as a NULL CDict when referenced into a CCtx.
584
585</p></pre><BR>
586
587<pre><b>size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); </b>/* Not ready yet ! */<b>
588</b><p> Add a prepared dictionary to cctx, it will used for next compression jobs.
589 Note that compression parameters will be enforced from within CDict.
590 Currently, they supercede any compression parameter previously set within CCtx.
591 The dictionary will remain valid for all future compression jobs performed using the same cctx.
592 @result : 0, or an error code (which can be tested with ZSTD_isError()).
593 Special : adding a NULL CDict means "return to no-dictionary mode".
594 Note 1 : Currently, only one dictionary can be managed.
595 Adding a new dictionary effectively "discards" any previous one.
596 Note 2 : CDict is just referenced, its lifetime must outlive CCtx.
597
598</p></pre><BR>
599
600<pre><b>typedef enum {
601 ZSTD_e_continue, </b>/* continue sending data, encoder transparently decides when to output result, depending on optimal conditions */<b>
602 ZSTD_e_flush, </b>/* flush any data provided and buffered so far - frame will continue, future data can still reference previous data for better compression */<b>
603 ZSTD_e_end </b>/* flush any remaining data and ends current frame. Any future compression starts a new frame. */<b>
604} ZSTD_EndDirective;
605</b></pre><BR>
606<pre><b>size_t ZSTD_compress_generic (ZSTD_CCtx* cctx,
607 void* dst, size_t dstCapacity, size_t* dstPos,
608 const void* src, size_t srcSize, size_t* srcPos,
609 ZSTD_EndDirective endOp);
610</b><p> Behave about the same as ZSTD_compressStream. To note :
611 - Compression parameters are pushed into CCtx before starting compression, using ZSTD_setCCtxParameter()
612 - Compression parameters cannot be changed once compression is started.
613 - *dstPos must be <= dstCapacity, *srcPos must be <= srcSize
614 - *dspPos and *srcPos will be updated. They are guaranteed to remain below their respective limit.
615 - @return provides the amount of data ready to flush and still within internal buffers
616 or an error code, which can be tested using ZSTD_isError().
617 if @return != 0, flush is not fully completed, so it must be called again to empty internal buffers.
618 - after a ZSTD_e_end directive, if internal buffer is not fully flushed,
619 only ZSTD_e_end and ZSTD_e_flush operations are allowed.
620 It is necessary to fully flush internal buffers
621 before changing compression parameters or start a new compression job.
622
623</p></pre><BR>
624
625<pre><b>size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx);
626</b><p> Return a CCtx to clean state.
627 Useful after an error, or to interrupt an ongoing compression job and start a new one.
628 It's allowed to change compression parameters after a reset.
629 Any internal data not yet flushed is cancelled.
630
631</p></pre><BR>
632
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]633<a name="Chapter15"></a><h2>Advanced decompression functions</h2><pre></pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]634
Przemyslaw Skibinski4da53212016-12-07 11:18:40 +0100[diff] [blame]635<pre><b>unsigned ZSTD_isFrame(const void* buffer, size_t size);
636</b><p> Tells if the content of `buffer` starts with a valid Frame Identifier.
637 Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
638 Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
639 Note 3 : Skippable Frame Identifiers are considered valid.
640</p></pre><BR>
641
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]642<pre><b>ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem);
643</b><p> Create a ZSTD decompression context using external alloc and free functions
644</p></pre><BR>
645
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]646<pre><b>ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize);
647</b><p> Create a digested dictionary, ready to start decompression operation without startup delay.
648 Dictionary content is simply referenced, and therefore stays in dictBuffer.
649 It is important that dictBuffer outlives DDict, it must remain read accessible throughout the lifetime of DDict
650</p></pre><BR>
651
Sean Purcelldec2b962017-03-14 11:24:09 -0700[diff] [blame]652<pre><b>ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
653 unsigned byReference, ZSTD_customMem customMem);
654</b><p> Create a ZSTD_DDict using external alloc and free, optionally by reference
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]655</p></pre><BR>
656
Przemyslaw Skibinski4da53212016-12-07 11:18:40 +0100[diff] [blame]657<pre><b>unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize);
658</b><p> Provides the dictID stored within dictionary.
659 if @return == 0, the dictionary is not conformant with Zstandard specification.
660 It can still be loaded, but as a content-only dictionary.
661</p></pre><BR>
662
663<pre><b>unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict);
664</b><p> Provides the dictID of the dictionary loaded into `ddict`.
665 If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
666 Non-conformant dictionaries can still be loaded, but as content-only dictionaries.
667</p></pre><BR>
668
669<pre><b>unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);
670</b><p> Provides the dictID required to decompressed the frame stored within `src`.
671 If @return == 0, the dictID could not be decoded.
672 This could for one of the following reasons :
673 - The frame does not require a dictionary to be decoded (most common case).
674 - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
675 Note : this use case also happens when using a non-conformant dictionary.
676 - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
677 - This is not a Zstandard frame.
Yann Collet202082f2017-04-28 16:56:39 -0700[diff] [blame]678 When identifying the exact failure cause, it's possible to use ZSTD_getFrameParams(), which will provide a more precise error code.
Przemyslaw Skibinski4da53212016-12-07 11:18:40 +0100[diff] [blame]679</p></pre><BR>
680
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]681<a name="Chapter16"></a><h2>Advanced streaming functions</h2><pre></pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]682
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]683<h3>Advanced Streaming compression functions</h3><pre></pre><b><pre>ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem);
684size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs, int compressionLevel, unsigned long long pledgedSrcSize); </b>/**< pledgedSrcSize must be correct, a size of 0 means unknown. for a frame size of 0 use initCStream_advanced */<b>
Yann Colleta1d67042017-05-08 17:51:49 -0700[diff] [blame]685size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs, const void* dict, size_t dictSize, int compressionLevel); </b>/**< note: a dict will not be used if dict == NULL or dictSize < 8. This result in the creation of an internal CDict */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]686size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs, const void* dict, size_t dictSize,
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]687 ZSTD_parameters params, unsigned long long pledgedSrcSize); </b>/**< pledgedSrcSize is optional and can be 0 (meaning unknown). note: if the contentSizeFlag is set, pledgedSrcSize == 0 means the source size is actually 0 */<b>
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]688size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict); </b>/**< note : cdict will just be referenced, and must outlive compression session */<b>
Yann Collet77bf59e2017-04-27 11:43:04 -0700[diff] [blame]689size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs, const ZSTD_CDict* cdict, unsigned long long pledgedSrcSize, ZSTD_frameParameters fParams); </b>/**< same as ZSTD_initCStream_usingCDict(), with control over frame parameters */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]690</pre></b><BR>
Yann Collet4b987ad2017-04-10 17:50:44 -0700[diff] [blame]691<pre><b>size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);
692</b><p> start a new compression job, using same parameters from previous job.
693 This is typically useful to skip dictionary loading stage, since it will re-use it in-place..
694 Note that zcs must be init at least once before using ZSTD_resetCStream().
695 pledgedSrcSize==0 means "srcSize unknown".
696 If pledgedSrcSize > 0, its value must be correct, as it will be written in header, and controlled at the end.
697 @return : 0, or an error code (which can be tested using ZSTD_isError())
698</p></pre><BR>
699
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]700<h3>Advanced Streaming decompression functions</h3><pre></pre><b><pre>typedef enum { DStream_p_maxWindowSize } ZSTD_DStreamParameter_e;
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]701ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]702size_t ZSTD_setDStreamParameter(ZSTD_DStream* zds, ZSTD_DStreamParameter_e paramType, unsigned paramValue);
Yann Collet5a36c062017-05-09 15:11:30 -0700[diff] [blame]703size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize); </b>/**< note: a dict will not be used if dict == NULL or dictSize < 8 */<b>
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]704size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict); </b>/**< note : ddict will just be referenced, and must outlive decompression session */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]705size_t ZSTD_resetDStream(ZSTD_DStream* zds); </b>/**< re-use decompression parameters from previous init; saves dictionary loading */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]706</pre></b><BR>
707<a name="Chapter17"></a><h2>Buffer-less and synchronous inner streaming functions</h2><pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]708 This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
709 But it's also a complex one, with many restrictions (documented below).
Przemyslaw Skibinski1fd5b452016-10-31 10:44:44 +0100[diff] [blame]710 Prefer using normal streaming API for an easier experience
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]711
712<BR></pre>
713
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]714<a name="Chapter18"></a><h2>Buffer-less streaming compression (synchronous mode)</h2><pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]715 A ZSTD_CCtx object is required to track streaming operations.
716 Use ZSTD_createCCtx() / ZSTD_freeCCtx() to manage resource.
717 ZSTD_CCtx object can be re-used multiple times within successive compression operations.
718
719 Start by initializing a context.
720 Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression,
721 or ZSTD_compressBegin_advanced(), for finer parameter control.
722 It's also possible to duplicate a reference context which has already been initialized, using ZSTD_copyCCtx()
723
724 Then, consume your input using ZSTD_compressContinue().
725 There are some important considerations to keep in mind when using this advanced function :
726 - ZSTD_compressContinue() has no internal buffer. It uses externally provided buffer only.
727 - Interface is synchronous : input is consumed entirely and produce 1+ (or more) compressed blocks.
728 - Caller must ensure there is enough space in `dst` to store compressed data under worst case scenario.
729 Worst case evaluation is provided by ZSTD_compressBound().
730 ZSTD_compressContinue() doesn't guarantee recover after a failed compression.
731 - ZSTD_compressContinue() presumes prior input ***is still accessible and unmodified*** (up to maximum distance size, see WindowLog).
732 It remembers all previous contiguous blocks, plus one separated memory segment (which can itself consists of multiple contiguous blocks)
733 - ZSTD_compressContinue() detects that prior input has been overwritten when `src` buffer overlaps.
734 In which case, it will "discard" the relevant memory section from its history.
735
736 Finish a frame with ZSTD_compressEnd(), which will write the last block(s) and optional checksum.
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]737 It's possible to use srcSize==0, in which case, it will write a final empty block to end the frame.
738 Without last block mark, frames will be considered unfinished (corrupted) by decoders.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]739
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]740 `ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress some new frame.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]741<BR></pre>
742
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]743<h3>Buffer-less streaming compression functions</h3><pre></pre><b><pre>size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]744size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]745size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize); </b>/**< pledgedSrcSize is optional and can be 0 (meaning unknown). note: if the contentSizeFlag is set, pledgedSrcSize == 0 means the source size is actually 0 */<b>
Yann Collet768df122017-04-26 15:42:10 -0700[diff] [blame]746size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); </b>/**< note: fails if cdict==NULL */<b>
Yann Collet715b9aa2017-04-18 13:55:53 -0700[diff] [blame]747size_t ZSTD_compressBegin_usingCDict_advanced(ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict, ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize); </b>/* compression parameters are already set within cdict. pledgedSrcSize=0 means null-size */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]748size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); </b>/**< note: if pledgedSrcSize can be 0, indicating unknown size. if it is non-zero, it must be accurate. for 0 size frames, use compressBegin_advanced */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]749</pre></b><BR>
750<a name="Chapter19"></a><h2>Buffer-less streaming decompression (synchronous mode)</h2><pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]751 A ZSTD_DCtx object is required to track streaming operations.
752 Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it.
753 A ZSTD_DCtx object can be re-used multiple times.
754
755 First typical operation is to retrieve frame parameters, using ZSTD_getFrameParams().
756 It fills a ZSTD_frameParams structure which provide important information to correctly decode the frame,
757 such as the minimum rolling buffer size to allocate to decompress data (`windowSize`),
758 and the dictionary ID used.
759 (Note : content size is optional, it may not be present. 0 means : content size unknown).
760 Note that these values could be wrong, either because of data malformation, or because an attacker is spoofing deliberate false information.
761 As a consequence, check that values remain within valid application range, especially `windowSize`, before allocation.
762 Each application can set its own limit, depending on local restrictions. For extended interoperability, it is recommended to support at least 8 MB.
763 Frame parameters are extracted from the beginning of the compressed frame.
764 Data fragment must be large enough to ensure successful decoding, typically `ZSTD_frameHeaderSize_max` bytes.
765 @result : 0 : successful decoding, the `ZSTD_frameParams` structure is correctly filled.
766 >0 : `srcSize` is too small, please provide at least @result bytes on next attempt.
767 errorCode, which can be tested using ZSTD_isError().
768
769 Start decompression, with ZSTD_decompressBegin() or ZSTD_decompressBegin_usingDict().
770 Alternatively, you can copy a prepared context, using ZSTD_copyDCtx().
771
772 Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue() alternatively.
773 ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize' to ZSTD_decompressContinue().
774 ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will fail.
775
776 @result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
777 It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some metadata item.
778 It can also be an error code, which can be tested with ZSTD_isError().
779
780 ZSTD_decompressContinue() needs previous data blocks during decompression, up to `windowSize`.
781 They should preferably be located contiguously, prior to current block.
782 Alternatively, a round buffer of sufficient size is also possible. Sufficient size is determined by frame parameters.
783 ZSTD_decompressContinue() is very sensitive to contiguity,
784 if 2 blocks don't follow each other, make sure that either the compressor breaks contiguity at the same place,
785 or that previous contiguous segment is large enough to properly handle maximum back-reference.
786
787 A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
788 Context can then be reset to start a new decompression.
789
790 Note : it's possible to know if next input to present is a header or a block, using ZSTD_nextInputType().
791 This information is not required to properly decode a frame.
792
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]793 == Special case : skippable frames
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]794
795 Skippable frames allow integration of user-defined data into a flow of concatenated frames.
796 Skippable frames will be ignored (skipped) by a decompressor. The format of skippable frames is as follows :
797 a) Skippable frame ID - 4 Bytes, Little endian format, any value from 0x184D2A50 to 0x184D2A5F
798 b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
799 c) Frame Content - any content (User Data) of length equal to Frame Size
800 For skippable frames ZSTD_decompressContinue() always returns 0.
801 For skippable frames ZSTD_getFrameParams() returns fparamsPtr->windowLog==0 what means that a frame is skippable.
Yann Collet831b4892017-02-23 23:09:10 -0800[diff] [blame]802 Note : If fparamsPtr->frameContentSize==0, it is ambiguous: the frame might actually be a Zstd encoded frame with no content.
803 For purposes of decompression, it is valid in both cases to skip the frame using
804 ZSTD_findFrameCompressedSize to find its size in bytes.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]805 It also returns Frame Size as fparamsPtr->frameContentSize.
806<BR></pre>
807
Yann Colletcef02d92017-05-10 11:14:08 -0700[diff] [blame]808<h3>Buffer-less streaming decompression functions</h3><pre></pre><b><pre>size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize); </b>/**< doesn't consume input, see details below */<b>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]809size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx);
810size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
811void ZSTD_copyDCtx(ZSTD_DCtx* dctx, const ZSTD_DCtx* preparedDCtx);
812size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx);
813size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
814typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
815ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]816</pre></b><BR>
817<a name="Chapter20"></a><h2>Block functions</h2><pre>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]818 Block functions produce and decode raw zstd blocks, without frame metadata.
819 Frame metadata cost is typically ~18 bytes, which can be non-negligible for very small blocks (< 100 bytes).
820 User will have to take in charge required information to regenerate data, such as compressed and content sizes.
821
822 A few rules to respect :
823 - Compressing and decompressing require a context structure
824 + Use ZSTD_createCCtx() and ZSTD_createDCtx()
825 - It is necessary to init context before starting
Yann Collet715b9aa2017-04-18 13:55:53 -0700[diff] [blame]826 + compression : any ZSTD_compressBegin*() variant, including with dictionary
827 + decompression : any ZSTD_decompressBegin*() variant, including with dictionary
828 + copyCCtx() and copyDCtx() can be used too
829 - Block size is limited, it must be <= ZSTD_getBlockSizeMax() <= ZSTD_BLOCKSIZE_ABSOLUTEMAX
830 + If input is larger than a block size, it's necessary to split input data into multiple blocks
831 + For inputs larger than a single block size, consider using the regular ZSTD_compress() instead.
832 Frame metadata is not that costly, and quickly becomes negligible as source size grows larger.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]833 - When a block is considered not compressible enough, ZSTD_compressBlock() result will be zero.
834 In which case, nothing is produced into `dst`.
835 + User must test for such outcome and deal directly with uncompressed data
836 + ZSTD_decompressBlock() doesn't accept uncompressed data as input !!!
Yann Collet715b9aa2017-04-18 13:55:53 -0700[diff] [blame]837 + In case of multiple successive blocks, should some of them be uncompressed,
838 decoder must be informed of their existence in order to follow proper history.
839 Use ZSTD_insertBlock() for such a case.
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]840<BR></pre>
841
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]842<h3>Raw zstd block functions</h3><pre></pre><b><pre>size_t ZSTD_getBlockSizeMax(ZSTD_CCtx* cctx);
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]843size_t ZSTD_compressBlock (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
844size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
845size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize); </b>/**< insert block into `dctx` history. Useful for uncompressed blocks */<b>
Yann Collet77575772017-02-22 01:10:43 -0800[diff] [blame]846</pre></b><BR>
Przemyslaw Skibinski86d94242016-10-24 16:07:53 +0200[diff] [blame]847</html>
848</body>