3 * \brief Data types and functions used in many places in liblzma API
9 * This file has been put into the public domain.
10 * You can do whatever you want with this file.
12 * See ../lzma.h for information about liblzma as a whole.
15 #ifndef LZMA_H_INTERNAL
16 # error Never include this file directly. Use <lzma.h> instead.
23 * This is here because C89 doesn't have stdbool.h. To set a value for
24 * variables having type lzma_bool, you can use
25 * - C99's `true' and `false' from stdbool.h;
26 * - C++'s internal `true' and `false'; or
27 * - integers one (true) and zero (false).
29 typedef unsigned char lzma_bool;
33 * \brief Type of reserved enumeration variable in structures
35 * To avoid breaking library ABI when new features are added, several
36 * structures contain extra variables that may be used in future. Since
37 * sizeof(enum) can be different than sizeof(int), and sizeof(enum) may
38 * even vary depending on the range of enumeration constants, we specify
39 * a separate type to be used for reserved enumeration variables. All
40 * enumeration constants in liblzma API will be non-negative and less
41 * than 128, which should guarantee that the ABI won't break even when
42 * new constants are added to existing enumerations.
45 LZMA_RESERVED_ENUM = 0
50 * \brief Return values used by several functions in liblzma
52 * Check the descriptions of specific functions to find out which return
53 * values they can return. With some functions the return values may have
54 * more specific meanings than described here; those differences are
55 * described per-function basis.
60 * \brief Operation completed successfully
65 * \brief End of stream was reached
67 * In encoder, LZMA_SYNC_FLUSH, LZMA_FULL_FLUSH, or
68 * LZMA_FINISH was finished. In decoder, this indicates
69 * that all the data was successfully decoded.
71 * In all cases, when LZMA_STREAM_END is returned, the last
72 * output bytes should be picked from strm->next_out.
77 * \brief Input stream has no integrity check
79 * This return value can be returned only if the
80 * LZMA_TELL_NO_CHECK flag was used when initializing
81 * the decoder. LZMA_NO_CHECK is just a warning, and
82 * the decoding can be continued normally.
84 * It is possible to call lzma_get_check() immediatelly after
85 * lzma_code has returned LZMA_NO_CHECK. The result will
86 * naturally be LZMA_CHECK_NONE, but the possibility to call
87 * lzma_get_check() may be convenient in some applications.
90 LZMA_UNSUPPORTED_CHECK = 3,
92 * \brief Cannot calculate the integrity check
94 * The usage of this return value is different in encoders
97 * Encoders can return this value only from the initialization
98 * function. If initialization fails with this value, the
99 * encoding cannot be done, because there's no way to produce
100 * output with the correct integrity check.
102 * Decoders can return this value only from lzma_code() and
103 * only if the LZMA_TELL_UNSUPPORTED_CHECK flag was used when
104 * initializing the decoder. The decoding can still be
105 * continued normally even if the check type is unsupported,
106 * but naturally the check will not be validated, and possible
107 * errors may go undetected.
109 * With decoder, it is possible to call lzma_get_check()
110 * immediatelly after lzma_code() has returned
111 * LZMA_UNSUPPORTED_CHECK. This way it is possible to find
112 * out what the unsupported Check ID was.
117 * \brief Integrity check type is now available
119 * This value can be returned only by the lzma_code() function
120 * and only if the decoder was initialized with the
121 * LZMA_TELL_ANY_CHECK flag. LZMA_GET_CHECK tells the
122 * application that it may now call lzma_get_check() to find
123 * out the Check ID. This can be used, for example, to
124 * implement a decoder that accepts only files that have
125 * strong enough integrity check.
130 * \brief Cannot allocate memory
132 * Memory allocation failed, or the size of the allocation
133 * would be greater than SIZE_MAX.
135 * Due to internal implementation reasons, the coding cannot
136 * be continued even if more memory were made available after
140 LZMA_MEMLIMIT_ERROR = 6,
142 * \brief Memory usage limit was reached
144 * Decoder would need more memory than allowed by the
145 * specified memory usage limit. To continue decoding,
146 * the memory usage limit has to be increased with
150 LZMA_FORMAT_ERROR = 7,
152 * \brief File format not recognized
154 * The decoder did not recognize the input as supported file
155 * format. This error can occur, for example, when trying to
156 * decode .lzma format file with lzma_stream_decoder,
157 * because lzma_stream_decoder accepts only the .xz format.
160 LZMA_OPTIONS_ERROR = 8,
162 * \brief Invalid or unsupported options
164 * Invalid or unsupported options, for example
165 * - unsupported filter(s) or filter options; or
166 * - reserved bits set in headers (decoder only).
168 * Rebuilding liblzma with more features enabled, or
169 * upgrading to a newer version of liblzma may help.
174 * \brief Data is corrupt
176 * The usage of this return value is different in encoders
177 * and decoders. In both encoder and decoder, the coding
178 * cannot continue after this error.
180 * Encoders return this if size limits of the target file
181 * format would be exceeded. These limits are huge, thus
182 * getting this error from an encoder is mostly theoretical.
183 * For example, the maximum compressed and uncompressed
184 * size of a .xz Stream created with lzma_stream_encoder is
185 * 2^63 - 1 bytes (one byte less than 8 EiB).
187 * Decoders return this error if the input data is corrupt.
188 * This can mean, for example, invalid CRC32 in headers
189 * or invalid check of uncompressed data.
194 * \brief No progress is possible
196 * This error code is returned when the coder cannot consume
197 * any new input and produce any new output. The most common
198 * reason for this error is that the input stream being
199 * decoded is truncated or corrupt.
201 * This error is not fatal. Coding can be continued normally
202 * by providing more input and/or more output space, if
205 * Typically the first call to lzma_code() that can do no
206 * progress returns LZMA_OK instead of LZMA_BUF_ERROR. Only
207 * the second consecutive call doing no progress will return
208 * LZMA_BUF_ERROR. This is intentional.
210 * With zlib, Z_BUF_ERROR may be returned even if the
211 * application is doing nothing wrong. The above hack
212 * guarantees that liblzma never returns LZMA_BUF_ERROR
213 * to properly written applications unless the input file
214 * is truncated or corrupt. This should simplify the
215 * applications a little.
218 LZMA_PROG_ERROR = 11,
220 * \brief Programming error
222 * This indicates that the arguments given to the function are
223 * invalid or the internal state of the decoder is corrupt.
224 * - Function arguments are invalid or the structures
225 * pointed by the argument pointers are invalid
226 * e.g. if strm->next_out has been set to NULL and
227 * strm->avail_out > 0 when calling lzma_code().
228 * - lzma_* functions have been called in wrong order
229 * e.g. lzma_code() was called right after lzma_end().
230 * - If errors occur randomly, the reason might be flaky
233 * If you think that your code is correct, this error code
234 * can be a sign of a bug in liblzma. See the documentation
235 * how to report bugs.
241 * \brief The `action' argument for lzma_code()
243 * After the first use of LZMA_SYNC_FLUSH, LZMA_FULL_FLUSH, or LZMA_FINISH,
244 * the same `action' must is used until lzma_code() returns LZMA_STREAM_END.
245 * Also, the amount of input (that is, strm->avail_in) must not be modified
246 * by the application until lzma_code() returns LZMA_STREAM_END. Changing the
247 * `action' or modifying the amount of input will make lzma_code() return
253 * \brief Continue coding
255 * Encoder: Encode as much input as possible. Some internal
256 * buffering will probably be done (depends on the filter
257 * chain in use), which causes latency: the input used won't
258 * usually be decodeable from the output of the same
261 * Decoder: Decode as much input as possible and produce as
262 * much output as possible.
267 * \brief Make all the input available at output
269 * Normally the encoder introduces some latency.
270 * LZMA_SYNC_FLUSH forces all the buffered data to be
271 * available at output without resetting the internal
272 * state of the encoder. This way it is possible to use
273 * compressed stream for example for communication over
276 * Only some filters support LZMA_SYNC_FLUSH. Trying to use
277 * LZMA_SYNC_FLUSH with filters that don't support it will
278 * make lzma_code() return LZMA_OPTIONS_ERROR. For example,
279 * LZMA1 doesn't support LZMA_SYNC_FLUSH but LZMA2 does.
281 * Using LZMA_SYNC_FLUSH very often can dramatically reduce
282 * the compression ratio. With some filters (for example,
283 * LZMA2), finetuning the compression options may help
284 * mitigate this problem significantly.
286 * Decoders don't support LZMA_SYNC_FLUSH.
291 * \brief Make all the input available at output
293 * Finish encoding of the current Block. All the input
294 * data going to the current Block must have been given
295 * to the encoder (the last bytes can still be pending in
296 * next_in). Call lzma_code() with LZMA_FULL_FLUSH until
297 * it returns LZMA_STREAM_END. Then continue normally with
298 * LZMA_RUN or finish the Stream with LZMA_FINISH.
300 * This action is currently supported only by Stream encoder
301 * and easy encoder (which uses Stream encoder). If there is
302 * no unfinished Block, no empty Block is created.
307 * \brief Finish the coding operation
309 * Finishes the coding operation. All the input data must
310 * have been given to the encoder (the last bytes can still
311 * be pending in next_in). Call lzma_code() with LZMA_FINISH
312 * until it returns LZMA_STREAM_END. Once LZMA_FINISH has
313 * been used, the amount of input must no longer be changed
314 * by the application.
316 * When decoding, using LZMA_FINISH is optional unless the
317 * LZMA_CONCATENATED flag was used when the decoder was
318 * initialized. When LZMA_CONCATENATED was not used, the only
319 * effect of LZMA_FINISH is that the amount of input must not
320 * be changed just like in the encoder.
326 * \brief Custom functions for memory handling
328 * A pointer to lzma_allocator may be passed via lzma_stream structure
329 * to liblzma, and some advanced functions take a pointer to lzma_allocator
330 * as a separate function argument. The library will use the functions
331 * specified in lzma_allocator for memory handling instead of the default
332 * malloc() and free().
334 * liblzma doesn't make an internal copy of lzma_allocator. Thus, it is
335 * OK to change these function pointers in the middle of the coding
336 * process, but obviously it must be done carefully to make sure that the
337 * replacement `free' can deallocate memory allocated by the earlier
338 * `alloc' function(s).
342 * \brief Pointer to a custom memory allocation function
344 * If you don't want a custom allocator, but still want
345 * custom free(), set this to NULL and liblzma will use
346 * the standard malloc().
348 * \param opaque lzma_allocator.opaque (see below)
349 * \param nmemb Number of elements like in calloc(). liblzma
350 * will always set nmemb to 1, so it is safe to
351 * ignore nmemb in a custom allocator if you like.
352 * The nmemb argument exists only for
353 * compatibility with zlib and libbzip2.
354 * \param size Size of an element in bytes.
355 * liblzma never sets this to zero.
357 * \return Pointer to the beginning of a memory block of
358 * `size' bytes, or NULL if allocation fails
359 * for some reason. When allocation fails, functions
360 * of liblzma return LZMA_MEM_ERROR.
362 * For performance reasons, the allocator should not waste time
363 * zeroing the allocated buffers. This is not only about speed, but
364 * also memory usage, since the operating system kernel doesn't
365 * necessarily allocate the requested memory in physical memory until
366 * it is actually used. With small input files liblzma may actually
367 * need only a fraction of the memory that it requested for allocation.
369 * \note LZMA_MEM_ERROR is also used when the size of the
370 * allocation would be greater than SIZE_MAX. Thus,
371 * don't assume that the custom allocator must have
372 * returned NULL if some function from liblzma
373 * returns LZMA_MEM_ERROR.
375 void *(LZMA_API_CALL *alloc)(void *opaque, size_t nmemb, size_t size);
378 * \brief Pointer to a custom memory freeing function
380 * If you don't want a custom freeing function, but still
381 * want a custom allocator, set this to NULL and liblzma
382 * will use the standard free().
384 * \param opaque lzma_allocator.opaque (see below)
385 * \param ptr Pointer returned by lzma_allocator.alloc(),
386 * or when it is set to NULL, a pointer returned
387 * by the standard malloc().
389 void (LZMA_API_CALL *free)(void *opaque, void *ptr);
392 * \brief Pointer passed to .alloc() and .free()
394 * opaque is passed as the first argument to lzma_allocator.alloc()
395 * and lzma_allocator.free(). This intended to ease implementing
396 * custom memory allocation functions for use with liblzma.
398 * If you don't need this, you should set this to NULL.
406 * \brief Internal data structure
408 * The contents of this structure is not visible outside the library.
410 typedef struct lzma_internal_s lzma_internal;
414 * \brief Passing data to and from liblzma
416 * The lzma_stream structure is used for
417 * - passing pointers to input and output buffers to liblzma;
418 * - defining custom memory hander functions; and
419 * - holding a pointer to coder-specific internal data structures.
423 * - After allocating lzma_stream (on stack or with malloc()), it must be
424 * initialized to LZMA_STREAM_INIT (see LZMA_STREAM_INIT for details).
426 * - Initialize a coder to the lzma_stream, for example by using
427 * lzma_easy_encoder() or lzma_auto_decoder(). Some notes:
428 * - In contrast to zlib, strm->next_in and strm->next_out are
429 * ignored by all initialization functions, thus it is safe
430 * to not initialize them yet.
431 * - The initialization functions always set strm->total_in and
432 * strm->total_out to zero.
433 * - If the initialization function fails, no memory is left allocated
434 * that would require freeing with lzma_end() even if some memory was
435 * associated with the lzma_stream structure when the initialization
436 * function was called.
438 * - Use lzma_code() to do the actual work.
440 * - Once the coding has been finished, the existing lzma_stream can be
441 * reused. It is OK to reuse lzma_stream with different initialization
442 * function without calling lzma_end() first. Old allocations are
443 * automatically freed.
445 * - Finally, use lzma_end() to free the allocated memory. lzma_end() never
446 * frees the lzma_stream structure itself.
448 * Application may modify the values of total_in and total_out as it wants.
449 * They are updated by liblzma to match the amount of data read and
450 * written, but aren't used for anything else.
453 const uint8_t *next_in; /**< Pointer to the next input byte. */
454 size_t avail_in; /**< Number of available input bytes in next_in. */
455 uint64_t total_in; /**< Total number of bytes read by liblzma. */
457 uint8_t *next_out; /**< Pointer to the next output position. */
458 size_t avail_out; /**< Amount of free space in next_out. */
459 uint64_t total_out; /**< Total number of bytes written by liblzma. */
462 * Custom memory allocation functions. Set to NULL to use
463 * the standard malloc() and free().
465 lzma_allocator *allocator;
467 /** Internal state is not visible to applications. */
468 lzma_internal *internal;
471 * Reserved space to allow possible future extensions without
472 * breaking the ABI. Excluding the initialization of this structure,
473 * you should not touch these, because the names of these variables
478 uint64_t reserved_int1;
479 uint64_t reserved_int2;
480 lzma_reserved_enum reserved_enum1;
481 lzma_reserved_enum reserved_enum2;
487 * \brief Initialization for lzma_stream
489 * When you declare an instance of lzma_stream, you can immediatelly
490 * initialize it so that initialization functions know that no memory
491 * has been allocated yet:
493 * lzma_stream strm = LZMA_STREAM_INIT;
495 * If you need to initialize a dynamically allocated lzma_stream, you can use
496 * memset(strm_pointer, 0, sizeof(lzma_stream)). Strictly speaking, this
497 * violates the C standard since NULL may have different internal
498 * representation than zero, but it should be portable enough in practice.
499 * Anyway, for maximum portability, you can use something like this:
501 * lzma_stream tmp = LZMA_STREAM_INIT;
504 #define LZMA_STREAM_INIT \
505 { NULL, 0, 0, NULL, 0, 0, NULL, NULL, \
506 NULL, NULL, 0, 0, LZMA_RESERVED_ENUM, LZMA_RESERVED_ENUM }
510 * \brief Encode or decode data
512 * Once the lzma_stream has been successfully initialized (e.g. with
513 * lzma_stream_encoder()), the actual encoding or decoding is done
514 * using this function. The application has to update strm->next_in,
515 * strm->avail_in, strm->next_out, and strm->avail_out to pass input
516 * to and get output from liblzma.
518 * See the description of the coder-specific initialization function to find
519 * out what `action' values are supported by the coder. See documentation of
520 * lzma_ret for the possible return values.
522 extern LZMA_API(lzma_ret) lzma_code(lzma_stream *strm, lzma_action action)
523 lzma_attr_warn_unused_result;
527 * \brief Free memory allocated for the coder data structures
529 * \param strm Pointer to lzma_stream that is at least initialized
530 * with LZMA_STREAM_INIT.
532 * After lzma_end(strm), strm->internal is guaranteed to be NULL. No other
533 * members of the lzma_stream structure are touched.
535 * \note zlib indicates an error if application end()s unfinished
536 * stream structure. liblzma doesn't do this, and assumes that
537 * application knows what it is doing.
539 extern LZMA_API(void) lzma_end(lzma_stream *strm);
543 * \brief Get the memory usage of decoder filter chain
545 * This function is currently supported only when *strm has been initialized
546 * with a function that takes a memlimit argument. With other functions, you
547 * should use e.g. lzma_raw_encoder_memusage() or lzma_raw_decoder_memusage()
548 * to estimate the memory requirements.
550 * This function is useful e.g. after LZMA_MEMLIMIT_ERROR to find out how big
551 * the memory usage limit should have been to decode the input. Note that
552 * this may give misleading information if decoding .xz Streams that have
553 * multiple Blocks, because each Block can have different memory requirements.
555 * \return Rough estimate of how much memory is currently allocated
556 * for the filter decoders. If no filter chain is currently
557 * allocated, some non-zero value is still returned, which is
558 * less than or equal to what any filter chain would indicate
559 * as its memory requirement.
561 * If this function isn't supported by *strm or some other error
562 * occurs, zero is returned.
564 extern LZMA_API(uint64_t) lzma_memusage(const lzma_stream *strm);
568 * \brief Get the current memory usage limit
570 * This function is supported only when *strm has been initialized with
571 * a function that takes a memlimit argument.
573 * \return On success, the current memory usage limit is returned
574 * (always non-zero). On error, zero is returned.
576 extern LZMA_API(uint64_t) lzma_memlimit_get(const lzma_stream *strm);
580 * \brief Set the memory usage limit
582 * This function is supported only when *strm has been initialized with
583 * a function that takes a memlimit argument.
585 * \return - LZMA_OK: New memory usage limit successfully set.
586 * - LZMA_MEMLIMIT_ERROR: The new limit is too small.
587 * The limit was not changed.
588 * - LZMA_PROG_ERROR: Invalid arguments, e.g. *strm doesn't
589 * support memory usage limit or memlimit was zero.
591 extern LZMA_API(lzma_ret) lzma_memlimit_set(
592 lzma_stream *strm, uint64_t memlimit);