1 | /* -*- mode: C; c-basic-offset: 4; indent-tabs-mode: nil -*- */ |
2 | // vim: expandtab:ts=8:sw=4:softtabstop=4: |
3 | /** |
4 | * \file lzma/block.h |
5 | * \brief .xz Block handling |
6 | */ |
7 | |
8 | /* |
9 | * Author: Lasse Collin |
10 | * |
11 | * This file has been put into the public domain. |
12 | * You can do whatever you want with this file. |
13 | * |
14 | * See ../lzma.h for information about liblzma as a whole. |
15 | */ |
16 | |
17 | #ifndef LZMA_H_INTERNAL |
18 | # error Never include this file directly. Use <lzma.h> instead. |
19 | #endif |
20 | |
21 | |
22 | /** |
23 | * \brief Options for the Block and Block Header encoders and decoders |
24 | * |
25 | * Different Block handling functions use different parts of this structure. |
26 | * Some read some members, other functions write, and some do both. Only the |
27 | * members listed for reading need to be initialized when the specified |
28 | * functions are called. The members marked for writing will be assigned |
29 | * new values at some point either by calling the given function or by |
30 | * later calls to lzma_code(). |
31 | */ |
32 | typedef struct { |
33 | /** |
34 | * \brief Block format version |
35 | * |
36 | * To prevent API and ABI breakages if new features are needed in |
37 | * Block, a version number is used to indicate which fields in this |
38 | * structure are in use. For now, version must always be zero. |
39 | * With non-zero version, most Block related functions will return |
40 | * LZMA_OPTIONS_ERROR. |
41 | * |
42 | * The decoding functions will always set this to the lowest value |
43 | * that supports all the features indicated by the Block Header field. |
44 | * The application must check that the version number set by the |
45 | * decoding functions is supported by the application. Otherwise it |
46 | * is possible that the application will decode the Block incorrectly. |
47 | * |
48 | * Read by: |
49 | * - lzma_block_header_size() |
50 | * - lzma_block_header_encode() |
51 | * - lzma_block_compressed_size() |
52 | * - lzma_block_unpadded_size() |
53 | * - lzma_block_total_size() |
54 | * - lzma_block_encoder() |
55 | * - lzma_block_decoder() |
56 | * - lzma_block_buffer_encode() |
57 | * - lzma_block_buffer_decode() |
58 | * |
59 | * Written by: |
60 | * - lzma_block_header_decode() |
61 | */ |
62 | uint32_t version; |
63 | |
64 | /** |
65 | * \brief Size of the Block Header field |
66 | * |
67 | * This is always a multiple of four. |
68 | * |
69 | * Read by: |
70 | * - lzma_block_header_encode() |
71 | * - lzma_block_header_decode() |
72 | * - lzma_block_compressed_size() |
73 | * - lzma_block_unpadded_size() |
74 | * - lzma_block_total_size() |
75 | * - lzma_block_decoder() |
76 | * - lzma_block_buffer_decode() |
77 | * |
78 | * Written by: |
79 | * - lzma_block_header_size() |
80 | * - lzma_block_buffer_encode() |
81 | */ |
82 | uint32_t ; |
83 | # define 8 |
84 | # define 1024 |
85 | |
86 | /** |
87 | * \brief Type of integrity Check |
88 | * |
89 | * The Check ID is not stored into the Block Header, thus its value |
90 | * must be provided also when decoding. |
91 | * |
92 | * Read by: |
93 | * - lzma_block_header_encode() |
94 | * - lzma_block_header_decode() |
95 | * - lzma_block_compressed_size() |
96 | * - lzma_block_unpadded_size() |
97 | * - lzma_block_total_size() |
98 | * - lzma_block_encoder() |
99 | * - lzma_block_decoder() |
100 | * - lzma_block_buffer_encode() |
101 | * - lzma_block_buffer_decode() |
102 | */ |
103 | lzma_check check; |
104 | |
105 | /** |
106 | * \brief Size of the Compressed Data in bytes |
107 | * |
108 | * Encoding: If this is not LZMA_VLI_UNKNOWN, Block Header encoder |
109 | * will store this value to the Block Header. Block encoder doesn't |
110 | * care about this value, but will set it once the encoding has been |
111 | * finished. |
112 | * |
113 | * Decoding: If this is not LZMA_VLI_UNKNOWN, Block decoder will |
114 | * verify that the size of the Compressed Data field matches |
115 | * compressed_size. |
116 | * |
117 | * Usually you don't know this value when encoding in streamed mode, |
118 | * and thus cannot write this field into the Block Header. |
119 | * |
120 | * In non-streamed mode you can reserve space for this field before |
121 | * encoding the actual Block. After encoding the data, finish the |
122 | * Block by encoding the Block Header. Steps in detail: |
123 | * |
124 | * - Set compressed_size to some big enough value. If you don't know |
125 | * better, use LZMA_VLI_MAX, but remember that bigger values take |
126 | * more space in Block Header. |
127 | * |
128 | * - Call lzma_block_header_size() to see how much space you need to |
129 | * reserve for the Block Header. |
130 | * |
131 | * - Encode the Block using lzma_block_encoder() and lzma_code(). |
132 | * It sets compressed_size to the correct value. |
133 | * |
134 | * - Use lzma_block_header_encode() to encode the Block Header. |
135 | * Because space was reserved in the first step, you don't need |
136 | * to call lzma_block_header_size() anymore, because due to |
137 | * reserving, header_size has to be big enough. If it is "too big", |
138 | * lzma_block_header_encode() will add enough Header Padding to |
139 | * make Block Header to match the size specified by header_size. |
140 | * |
141 | * Read by: |
142 | * - lzma_block_header_size() |
143 | * - lzma_block_header_encode() |
144 | * - lzma_block_compressed_size() |
145 | * - lzma_block_unpadded_size() |
146 | * - lzma_block_total_size() |
147 | * - lzma_block_decoder() |
148 | * - lzma_block_buffer_decode() |
149 | * |
150 | * Written by: |
151 | * - lzma_block_header_decode() |
152 | * - lzma_block_compressed_size() |
153 | * - lzma_block_encoder() |
154 | * - lzma_block_decoder() |
155 | * - lzma_block_buffer_encode() |
156 | * - lzma_block_buffer_decode() |
157 | */ |
158 | lzma_vli compressed_size; |
159 | |
160 | /** |
161 | * \brief Uncompressed Size in bytes |
162 | * |
163 | * This is handled very similarly to compressed_size above. |
164 | * |
165 | * uncompressed_size is needed by fewer functions than |
166 | * compressed_size. This is because uncompressed_size isn't |
167 | * needed to validate that Block stays within proper limits. |
168 | * |
169 | * Read by: |
170 | * - lzma_block_header_size() |
171 | * - lzma_block_header_encode() |
172 | * - lzma_block_decoder() |
173 | * - lzma_block_buffer_decode() |
174 | * |
175 | * Written by: |
176 | * - lzma_block_header_decode() |
177 | * - lzma_block_encoder() |
178 | * - lzma_block_decoder() |
179 | * - lzma_block_buffer_encode() |
180 | * - lzma_block_buffer_decode() |
181 | */ |
182 | lzma_vli uncompressed_size; |
183 | |
184 | /** |
185 | * \brief Array of filters |
186 | * |
187 | * There can be 1-4 filters. The end of the array is marked with |
188 | * .id = LZMA_VLI_UNKNOWN. |
189 | * |
190 | * Read by: |
191 | * - lzma_block_header_size() |
192 | * - lzma_block_header_encode() |
193 | * - lzma_block_encoder() |
194 | * - lzma_block_decoder() |
195 | * - lzma_block_buffer_encode() |
196 | * - lzma_block_buffer_decode() |
197 | * |
198 | * Written by: |
199 | * - lzma_block_header_decode(): Note that this does NOT free() |
200 | * the old filter options structures. All unused filters[] will |
201 | * have .id == LZMA_VLI_UNKNOWN and .options == NULL. If |
202 | * decoding fails, all filters[] are guaranteed to be |
203 | * LZMA_VLI_UNKNOWN and NULL. |
204 | * |
205 | * \note Because of the array is terminated with |
206 | * .id = LZMA_VLI_UNKNOWN, the actual array must |
207 | * have LZMA_FILTERS_MAX + 1 members or the Block |
208 | * Header decoder will overflow the buffer. |
209 | */ |
210 | lzma_filter *filters; |
211 | |
212 | /** |
213 | * \brief Raw value stored in the Check field |
214 | * |
215 | * After successful coding, the first lzma_check_size(check) bytes |
216 | * of this array contain the raw value stored in the Check field. |
217 | * |
218 | * Note that CRC32 and CRC64 are stored in little endian byte order. |
219 | * Take it into account if you display the Check values to the user. |
220 | * |
221 | * Written by: |
222 | * - lzma_block_encoder() |
223 | * - lzma_block_decoder() |
224 | * - lzma_block_buffer_encode() |
225 | * - lzma_block_buffer_decode() |
226 | */ |
227 | uint8_t raw_check[LZMA_CHECK_SIZE_MAX]; |
228 | |
229 | /* |
230 | * Reserved space to allow possible future extensions without |
231 | * breaking the ABI. You should not touch these, because the names |
232 | * of these variables may change. These are and will never be used |
233 | * with the currently supported options, so it is safe to leave these |
234 | * uninitialized. |
235 | */ |
236 | void *reserved_ptr1; |
237 | void *reserved_ptr2; |
238 | void *reserved_ptr3; |
239 | uint32_t reserved_int1; |
240 | uint32_t reserved_int2; |
241 | lzma_vli reserved_int3; |
242 | lzma_vli reserved_int4; |
243 | lzma_vli reserved_int5; |
244 | lzma_vli reserved_int6; |
245 | lzma_vli reserved_int7; |
246 | lzma_vli reserved_int8; |
247 | lzma_reserved_enum reserved_enum1; |
248 | lzma_reserved_enum reserved_enum2; |
249 | lzma_reserved_enum reserved_enum3; |
250 | lzma_reserved_enum reserved_enum4; |
251 | lzma_bool reserved_bool1; |
252 | lzma_bool reserved_bool2; |
253 | lzma_bool reserved_bool3; |
254 | lzma_bool reserved_bool4; |
255 | lzma_bool reserved_bool5; |
256 | lzma_bool reserved_bool6; |
257 | lzma_bool reserved_bool7; |
258 | lzma_bool reserved_bool8; |
259 | |
260 | } lzma_block; |
261 | |
262 | |
263 | /** |
264 | * \brief Decode the Block Header Size field |
265 | * |
266 | * To decode Block Header using lzma_block_header_decode(), the size of the |
267 | * Block Header has to be known and stored into lzma_block.header_size. |
268 | * The size can be calculated from the first byte of a Block using this macro. |
269 | * Note that if the first byte is 0x00, it indicates beginning of Index; use |
270 | * this macro only when the byte is not 0x00. |
271 | * |
272 | * There is no encoding macro, because Block Header encoder is enough for that. |
273 | */ |
274 | #define (b) (((uint32_t)(b) + 1) * 4) |
275 | |
276 | |
277 | /** |
278 | * \brief Calculate Block Header Size |
279 | * |
280 | * Calculate the minimum size needed for the Block Header field using the |
281 | * settings specified in the lzma_block structure. Note that it is OK to |
282 | * increase the calculated header_size value as long as it is a multiple of |
283 | * four and doesn't exceed LZMA_BLOCK_HEADER_SIZE_MAX. Increasing header_size |
284 | * just means that lzma_block_header_encode() will add Header Padding. |
285 | * |
286 | * \return - LZMA_OK: Size calculated successfully and stored to |
287 | * block->header_size. |
288 | * - LZMA_OPTIONS_ERROR: Unsupported version, filters or |
289 | * filter options. |
290 | * - LZMA_PROG_ERROR: Invalid values like compressed_size == 0. |
291 | * |
292 | * \note This doesn't check that all the options are valid i.e. this |
293 | * may return LZMA_OK even if lzma_block_header_encode() or |
294 | * lzma_block_encoder() would fail. If you want to validate the |
295 | * filter chain, consider using lzma_memlimit_encoder() which as |
296 | * a side-effect validates the filter chain. |
297 | */ |
298 | extern LZMA_API(lzma_ret) (lzma_block *block) |
299 | lzma_nothrow lzma_attr_warn_unused_result; |
300 | |
301 | |
302 | /** |
303 | * \brief Encode Block Header |
304 | * |
305 | * The caller must have calculated the size of the Block Header already with |
306 | * lzma_block_header_size(). If a value larger than the one calculated by |
307 | * lzma_block_header_size() is used, the Block Header will be padded to the |
308 | * specified size. |
309 | * |
310 | * \param out Beginning of the output buffer. This must be |
311 | * at least block->header_size bytes. |
312 | * \param block Block options to be encoded. |
313 | * |
314 | * \return - LZMA_OK: Encoding was successful. block->header_size |
315 | * bytes were written to output buffer. |
316 | * - LZMA_OPTIONS_ERROR: Invalid or unsupported options. |
317 | * - LZMA_PROG_ERROR: Invalid arguments, for example |
318 | * block->header_size is invalid or block->filters is NULL. |
319 | */ |
320 | extern LZMA_API(lzma_ret) ( |
321 | const lzma_block *block, uint8_t *out) |
322 | lzma_nothrow lzma_attr_warn_unused_result; |
323 | |
324 | |
325 | /** |
326 | * \brief Decode Block Header |
327 | * |
328 | * The size of the Block Header must have already been decoded with |
329 | * lzma_block_header_size_decode() macro and stored to block->header_size. |
330 | * block->filters must have been allocated, but not necessarily initialized. |
331 | * Possible existing filter options are _not_ freed. |
332 | * |
333 | * \param block Destination for block options with header_size |
334 | * properly initialized. |
335 | * \param allocator lzma_allocator for custom allocator functions. |
336 | * Set to NULL to use malloc() (and also free() |
337 | * if an error occurs). |
338 | * \param in Beginning of the input buffer. This must be |
339 | * at least block->header_size bytes. |
340 | * |
341 | * \return - LZMA_OK: Decoding was successful. block->header_size |
342 | * bytes were read from the input buffer. |
343 | * - LZMA_OPTIONS_ERROR: The Block Header specifies some |
344 | * unsupported options such as unsupported filters. |
345 | * - LZMA_DATA_ERROR: Block Header is corrupt, for example, |
346 | * the CRC32 doesn't match. |
347 | * - LZMA_PROG_ERROR: Invalid arguments, for example |
348 | * block->header_size is invalid or block->filters is NULL. |
349 | */ |
350 | extern LZMA_API(lzma_ret) (lzma_block *block, |
351 | lzma_allocator *allocator, const uint8_t *in) |
352 | lzma_nothrow lzma_attr_warn_unused_result; |
353 | |
354 | |
355 | /** |
356 | * \brief Validate and set Compressed Size according to Unpadded Size |
357 | * |
358 | * Block Header stores Compressed Size, but Index has Unpadded Size. If the |
359 | * application has already parsed the Index and is now decoding Blocks, |
360 | * it can calculate Compressed Size from Unpadded Size. This function does |
361 | * exactly that with error checking: |
362 | * |
363 | * - Compressed Size calculated from Unpadded Size must be positive integer, |
364 | * that is, Unpadded Size must be big enough that after Block Header and |
365 | * Check fields there's still at least one byte for Compressed Size. |
366 | * |
367 | * - If Compressed Size was present in Block Header, the new value |
368 | * calculated from Unpadded Size is compared against the value |
369 | * from Block Header. |
370 | * |
371 | * \note This function must be called _after_ decoding the Block Header |
372 | * field so that it can properly validate Compressed Size if it |
373 | * was present in Block Header. |
374 | * |
375 | * \return - LZMA_OK: block->compressed_size was set successfully. |
376 | * - LZMA_DATA_ERROR: unpadded_size is too small compared to |
377 | * block->header_size and lzma_check_size(block->check). |
378 | * - LZMA_PROG_ERROR: Some values are invalid. For example, |
379 | * block->header_size must be a multiple of four and |
380 | * between 8 and 1024 inclusive. |
381 | */ |
382 | extern LZMA_API(lzma_ret) lzma_block_compressed_size( |
383 | lzma_block *block, lzma_vli unpadded_size) |
384 | lzma_nothrow lzma_attr_warn_unused_result; |
385 | |
386 | |
387 | /** |
388 | * \brief Calculate Unpadded Size |
389 | * |
390 | * The Index field stores Unpadded Size and Uncompressed Size. The latter |
391 | * can be taken directly from the lzma_block structure after coding a Block, |
392 | * but Unpadded Size needs to be calculated from Block Header Size, |
393 | * Compressed Size, and size of the Check field. This is where this function |
394 | * is needed. |
395 | * |
396 | * \return Unpadded Size on success, or zero on error. |
397 | */ |
398 | extern LZMA_API(lzma_vli) lzma_block_unpadded_size(const lzma_block *block) |
399 | lzma_nothrow lzma_attr_pure; |
400 | |
401 | |
402 | /** |
403 | * \brief Calculate the total encoded size of a Block |
404 | * |
405 | * This is equivalent to lzma_block_unpadded_size() except that the returned |
406 | * value includes the size of the Block Padding field. |
407 | * |
408 | * \return On success, total encoded size of the Block. On error, |
409 | * zero is returned. |
410 | */ |
411 | extern LZMA_API(lzma_vli) lzma_block_total_size(const lzma_block *block) |
412 | lzma_nothrow lzma_attr_pure; |
413 | |
414 | |
415 | /** |
416 | * \brief Initialize .xz Block encoder |
417 | * |
418 | * Valid actions for lzma_code() are LZMA_RUN, LZMA_SYNC_FLUSH (only if the |
419 | * filter chain supports it), and LZMA_FINISH. |
420 | * |
421 | * \return - LZMA_OK: All good, continue with lzma_code(). |
422 | * - LZMA_MEM_ERROR |
423 | * - LZMA_OPTIONS_ERROR |
424 | * - LZMA_UNSUPPORTED_CHECK: block->check specfies a Check ID |
425 | * that is not supported by this buid of liblzma. Initializing |
426 | * the encoder failed. |
427 | * - LZMA_PROG_ERROR |
428 | */ |
429 | extern LZMA_API(lzma_ret) lzma_block_encoder( |
430 | lzma_stream *strm, lzma_block *block) |
431 | lzma_nothrow lzma_attr_warn_unused_result; |
432 | |
433 | |
434 | /** |
435 | * \brief Initialize .xz Block decoder |
436 | * |
437 | * Valid actions for lzma_code() are LZMA_RUN and LZMA_FINISH. Using |
438 | * LZMA_FINISH is not required. It is supported only for convenience. |
439 | * |
440 | * \return - LZMA_OK: All good, continue with lzma_code(). |
441 | * - LZMA_UNSUPPORTED_CHECK: Initialization was successful, but |
442 | * the given Check ID is not supported, thus Check will be |
443 | * ignored. |
444 | * - LZMA_PROG_ERROR |
445 | * - LZMA_MEM_ERROR |
446 | */ |
447 | extern LZMA_API(lzma_ret) lzma_block_decoder( |
448 | lzma_stream *strm, lzma_block *block) |
449 | lzma_nothrow lzma_attr_warn_unused_result; |
450 | |
451 | |
452 | /** |
453 | * \brief Calculate maximum output size for single-call Block encoding |
454 | * |
455 | * This is equivalent to lzma_stream_buffer_bound() but for .xz Blocks. |
456 | * See the documentation of lzma_stream_buffer_bound(). |
457 | */ |
458 | extern LZMA_API(size_t) lzma_block_buffer_bound(size_t uncompressed_size) |
459 | lzma_nothrow; |
460 | |
461 | |
462 | /** |
463 | * \brief Single-call .xz Block encoder |
464 | * |
465 | * In contrast to the multi-call encoder initialized with |
466 | * lzma_block_encoder(), this function encodes also the Block Header. This |
467 | * is required to make it possible to write appropriate Block Header also |
468 | * in case the data isn't compressible, and different filter chain has to be |
469 | * used to encode the data in uncompressed form using uncompressed chunks |
470 | * of the LZMA2 filter. |
471 | * |
472 | * When the data isn't compressible, header_size, compressed_size, and |
473 | * uncompressed_size are set just like when the data was compressible, but |
474 | * it is possible that header_size is too small to hold the filter chain |
475 | * specified in block->filters, because that isn't necessarily the filter |
476 | * chain that was actually used to encode the data. lzma_block_unpadded_size() |
477 | * still works normally, because it doesn't read the filters array. |
478 | * |
479 | * \param block Block options: block->version, block->check, |
480 | * and block->filters must have been initialized. |
481 | * \param allocator lzma_allocator for custom allocator functions. |
482 | * Set to NULL to use malloc() and free(). |
483 | * \param in Beginning of the input buffer |
484 | * \param in_size Size of the input buffer |
485 | * \param out Beginning of the output buffer |
486 | * \param out_pos The next byte will be written to out[*out_pos]. |
487 | * *out_pos is updated only if encoding succeeds. |
488 | * \param out_size Size of the out buffer; the first byte into |
489 | * which no data is written to is out[out_size]. |
490 | * |
491 | * \return - LZMA_OK: Encoding was successful. |
492 | * - LZMA_BUF_ERROR: Not enough output buffer space. |
493 | * - LZMA_OPTIONS_ERROR |
494 | * - LZMA_MEM_ERROR |
495 | * - LZMA_DATA_ERROR |
496 | * - LZMA_PROG_ERROR |
497 | */ |
498 | extern LZMA_API(lzma_ret) lzma_block_buffer_encode( |
499 | lzma_block *block, lzma_allocator *allocator, |
500 | const uint8_t *in, size_t in_size, |
501 | uint8_t *out, size_t *out_pos, size_t out_size) |
502 | lzma_nothrow lzma_attr_warn_unused_result; |
503 | |
504 | |
505 | /** |
506 | * \brief Single-call .xz Block decoder |
507 | * |
508 | * This is single-call equivalent of lzma_block_decoder(), and requires that |
509 | * the caller has already decoded Block Header and checked its memory usage. |
510 | * |
511 | * \param block Block options just like with lzma_block_decoder(). |
512 | * \param allocator lzma_allocator for custom allocator functions. |
513 | * Set to NULL to use malloc() and free(). |
514 | * \param in Beginning of the input buffer |
515 | * \param in_pos The next byte will be read from in[*in_pos]. |
516 | * *in_pos is updated only if decoding succeeds. |
517 | * \param in_size Size of the input buffer; the first byte that |
518 | * won't be read is in[in_size]. |
519 | * \param out Beginning of the output buffer |
520 | * \param out_pos The next byte will be written to out[*out_pos]. |
521 | * *out_pos is updated only if encoding succeeds. |
522 | * \param out_size Size of the out buffer; the first byte into |
523 | * which no data is written to is out[out_size]. |
524 | * |
525 | * \return - LZMA_OK: Decoding was successful. |
526 | * - LZMA_OPTIONS_ERROR |
527 | * - LZMA_DATA_ERROR |
528 | * - LZMA_MEM_ERROR |
529 | * - LZMA_BUF_ERROR: Output buffer was too small. |
530 | * - LZMA_PROG_ERROR |
531 | */ |
532 | extern LZMA_API(lzma_ret) lzma_block_buffer_decode( |
533 | lzma_block *block, lzma_allocator *allocator, |
534 | const uint8_t *in, size_t *in_pos, size_t in_size, |
535 | uint8_t *out, size_t *out_pos, size_t out_size) |
536 | lzma_nothrow; |
537 | |