archive_write_add_filter_xz.c revision 231200
1/*- 2 * Copyright (c) 2009,2010 Michihiro NAKAJIMA 3 * Copyright (c) 2003-2010 Tim Kientzle 4 * All rights reserved. 5 * 6 * Redistribution and use in source and binary forms, with or without 7 * modification, are permitted provided that the following conditions 8 * are met: 9 * 1. Redistributions of source code must retain the above copyright 10 * notice, this list of conditions and the following disclaimer. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR 16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 18 * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, 19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 25 */ 26 27#include "archive_platform.h" 28 29__FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_xz.c 201108 2009-12-28 03:28:21Z kientzle $"); 30 31#ifdef HAVE_ERRNO_H 32#include <errno.h> 33#endif 34#ifdef HAVE_STDLIB_H 35#include <stdlib.h> 36#endif 37#ifdef HAVE_STRING_H 38#include <string.h> 39#endif 40#include <time.h> 41#ifdef HAVE_LZMA_H 42#include <lzma.h> 43#endif 44 45#include "archive.h" 46#include "archive_endian.h" 47#include "archive_private.h" 48#include "archive_write_private.h" 49 50#if ARCHIVE_VERSION_NUMBER < 4000000 51int 52archive_write_set_compression_lzip(struct archive *a) 53{ 54 __archive_write_filters_free(a); 55 return (archive_write_add_filter_lzip(a)); 56} 57 58int 59archive_write_set_compression_lzma(struct archive *a) 60{ 61 __archive_write_filters_free(a); 62 return (archive_write_add_filter_lzma(a)); 63} 64 65int 66archive_write_set_compression_xz(struct archive *a) 67{ 68 __archive_write_filters_free(a); 69 return (archive_write_add_filter_xz(a)); 70} 71 72#endif 73 74#ifndef HAVE_LZMA_H 75int 76archive_write_add_filter_xz(struct archive *a) 77{ 78 archive_set_error(a, ARCHIVE_ERRNO_MISC, 79 "xz compression not supported on this platform"); 80 return (ARCHIVE_FATAL); 81} 82 83int 84archive_write_add_filter_lzma(struct archive *a) 85{ 86 archive_set_error(a, ARCHIVE_ERRNO_MISC, 87 "lzma compression not supported on this platform"); 88 return (ARCHIVE_FATAL); 89} 90 91int 92archive_write_add_filter_lzip(struct archive *a) 93{ 94 archive_set_error(a, ARCHIVE_ERRNO_MISC, 95 "lzma compression not supported on this platform"); 96 return (ARCHIVE_FATAL); 97} 98#else 99/* Don't compile this if we don't have liblzma. */ 100 101struct private_data { 102 int compression_level; 103 lzma_stream stream; 104 lzma_filter lzmafilters[2]; 105 lzma_options_lzma lzma_opt; 106 int64_t total_in; 107 unsigned char *compressed; 108 size_t compressed_buffer_size; 109 int64_t total_out; 110 /* the CRC32 value of uncompressed data for lzip */ 111 uint32_t crc32; 112}; 113 114static int archive_compressor_xz_options(struct archive_write_filter *, 115 const char *, const char *); 116static int archive_compressor_xz_open(struct archive_write_filter *); 117static int archive_compressor_xz_write(struct archive_write_filter *, 118 const void *, size_t); 119static int archive_compressor_xz_close(struct archive_write_filter *); 120static int archive_compressor_xz_free(struct archive_write_filter *); 121static int drive_compressor(struct archive_write_filter *, 122 struct private_data *, int finishing); 123 124struct option_value { 125 uint32_t dict_size; 126 uint32_t nice_len; 127 lzma_match_finder mf; 128}; 129static const struct option_value option_values[] = { 130 { 1 << 16, 32, LZMA_MF_HC3}, 131 { 1 << 20, 32, LZMA_MF_HC3}, 132 { 3 << 19, 32, LZMA_MF_HC4}, 133 { 1 << 21, 32, LZMA_MF_BT4}, 134 { 3 << 20, 32, LZMA_MF_BT4}, 135 { 1 << 22, 32, LZMA_MF_BT4}, 136 { 1 << 23, 64, LZMA_MF_BT4}, 137 { 1 << 24, 64, LZMA_MF_BT4}, 138 { 3 << 23, 64, LZMA_MF_BT4}, 139 { 1 << 25, 64, LZMA_MF_BT4} 140}; 141 142static int 143common_setup(struct archive_write_filter *f) 144{ 145 struct private_data *data; 146 struct archive_write *a = (struct archive_write *)f->archive; 147 data = calloc(1, sizeof(*data)); 148 if (data == NULL) { 149 archive_set_error(&a->archive, ENOMEM, "Out of memory"); 150 return (ARCHIVE_FATAL); 151 } 152 f->data = data; 153 data->compression_level = LZMA_PRESET_DEFAULT; 154 f->open = &archive_compressor_xz_open; 155 f->close = archive_compressor_xz_close; 156 f->free = archive_compressor_xz_free; 157 f->options = &archive_compressor_xz_options; 158 return (ARCHIVE_OK); 159} 160 161/* 162 * Add an xz compression filter to this write handle. 163 */ 164int 165archive_write_add_filter_xz(struct archive *_a) 166{ 167 struct archive_write_filter *f; 168 int r; 169 170 archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, 171 ARCHIVE_STATE_NEW, "archive_write_add_filter_xz"); 172 f = __archive_write_allocate_filter(_a); 173 r = common_setup(f); 174 if (r == ARCHIVE_OK) { 175 f->code = ARCHIVE_COMPRESSION_XZ; 176 f->name = "xz"; 177 } 178 return (r); 179} 180 181/* LZMA is handled identically, we just need a different compression 182 * code set. (The liblzma setup looks at the code to determine 183 * the one place that XZ and LZMA require different handling.) */ 184int 185archive_write_add_filter_lzma(struct archive *_a) 186{ 187 struct archive_write_filter *f; 188 int r; 189 190 archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, 191 ARCHIVE_STATE_NEW, "archive_write_add_filter_lzma"); 192 f = __archive_write_allocate_filter(_a); 193 r = common_setup(f); 194 if (r == ARCHIVE_OK) { 195 f->code = ARCHIVE_COMPRESSION_LZMA; 196 f->name = "lzma"; 197 } 198 return (r); 199} 200 201int 202archive_write_add_filter_lzip(struct archive *_a) 203{ 204 struct archive_write_filter *f; 205 int r; 206 207 archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, 208 ARCHIVE_STATE_NEW, "archive_write_add_filter_lzip"); 209 f = __archive_write_allocate_filter(_a); 210 r = common_setup(f); 211 if (r == ARCHIVE_OK) { 212 f->code = ARCHIVE_COMPRESSION_LZIP; 213 f->name = "lzip"; 214 } 215 return (r); 216} 217 218static int 219archive_compressor_xz_init_stream(struct archive_write_filter *f, 220 struct private_data *data) 221{ 222 static const lzma_stream lzma_stream_init_data = LZMA_STREAM_INIT; 223 int ret; 224 225 data->stream = lzma_stream_init_data; 226 data->stream.next_out = data->compressed; 227 data->stream.avail_out = data->compressed_buffer_size; 228 if (f->code == ARCHIVE_COMPRESSION_XZ) 229 ret = lzma_stream_encoder(&(data->stream), 230 data->lzmafilters, LZMA_CHECK_CRC64); 231 else if (f->code == ARCHIVE_COMPRESSION_LZMA) 232 ret = lzma_alone_encoder(&(data->stream), &data->lzma_opt); 233 else { /* ARCHIVE_COMPRESSION_LZIP */ 234 int dict_size = data->lzma_opt.dict_size; 235 int ds, log2dic, wedges; 236 237 /* Calculate a coded dictionary size */ 238 if (dict_size < (1 << 12) || dict_size > (1 << 27)) { 239 archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, 240 "Unacceptable dictionary dize for lzip: %d", 241 dict_size); 242 return (ARCHIVE_FATAL); 243 } 244 for (log2dic = 27; log2dic >= 12; log2dic--) { 245 if (dict_size & (1 << log2dic)) 246 break; 247 } 248 if (dict_size > (1 << log2dic)) { 249 log2dic++; 250 wedges = 251 ((1 << log2dic) - dict_size) / (1 << (log2dic - 4)); 252 } else 253 wedges = 0; 254 ds = ((wedges << 5) & 0xe0) | (log2dic & 0x1f); 255 256 data->crc32 = 0; 257 /* Make a header */ 258 data->compressed[0] = 0x4C; 259 data->compressed[1] = 0x5A; 260 data->compressed[2] = 0x49; 261 data->compressed[3] = 0x50; 262 data->compressed[4] = 1;/* Version */ 263 data->compressed[5] = (unsigned char)ds; 264 data->stream.next_out += 6; 265 data->stream.avail_out -= 6; 266 267 ret = lzma_raw_encoder(&(data->stream), data->lzmafilters); 268 } 269 if (ret == LZMA_OK) 270 return (ARCHIVE_OK); 271 272 switch (ret) { 273 case LZMA_MEM_ERROR: 274 archive_set_error(f->archive, ENOMEM, 275 "Internal error initializing compression library: " 276 "Cannot allocate memory"); 277 break; 278 default: 279 archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, 280 "Internal error initializing compression library: " 281 "It's a bug in liblzma"); 282 break; 283 } 284 return (ARCHIVE_FATAL); 285} 286 287/* 288 * Setup callback. 289 */ 290static int 291archive_compressor_xz_open(struct archive_write_filter *f) 292{ 293 struct private_data *data = f->data; 294 int ret; 295 296 ret = __archive_write_open_filter(f->next_filter); 297 if (ret != ARCHIVE_OK) 298 return (ret); 299 300 if (data->compressed == NULL) { 301 data->compressed_buffer_size = 65536; 302 data->compressed 303 = (unsigned char *)malloc(data->compressed_buffer_size); 304 if (data->compressed == NULL) { 305 archive_set_error(f->archive, ENOMEM, 306 "Can't allocate data for compression buffer"); 307 return (ARCHIVE_FATAL); 308 } 309 } 310 311 f->write = archive_compressor_xz_write; 312 313 /* Initialize compression library. */ 314 if (f->code == ARCHIVE_COMPRESSION_LZIP) { 315 const struct option_value *val = 316 &option_values[data->compression_level]; 317 318 data->lzma_opt.dict_size = val->dict_size; 319 data->lzma_opt.preset_dict = NULL; 320 data->lzma_opt.preset_dict_size = 0; 321 data->lzma_opt.lc = LZMA_LC_DEFAULT; 322 data->lzma_opt.lp = LZMA_LP_DEFAULT; 323 data->lzma_opt.pb = LZMA_PB_DEFAULT; 324 data->lzma_opt.mode = 325 data->compression_level<= 2? LZMA_MODE_FAST:LZMA_MODE_NORMAL; 326 data->lzma_opt.nice_len = val->nice_len; 327 data->lzma_opt.mf = val->mf; 328 data->lzma_opt.depth = 0; 329 data->lzmafilters[0].id = LZMA_FILTER_LZMA1; 330 data->lzmafilters[0].options = &data->lzma_opt; 331 data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ 332 } else { 333 if (lzma_lzma_preset(&data->lzma_opt, data->compression_level)) { 334 archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, 335 "Internal error initializing compression library"); 336 } 337 data->lzmafilters[0].id = LZMA_FILTER_LZMA2; 338 data->lzmafilters[0].options = &data->lzma_opt; 339 data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ 340 } 341 ret = archive_compressor_xz_init_stream(f, data); 342 if (ret == LZMA_OK) { 343 f->data = data; 344 return (0); 345 } 346 return (ARCHIVE_FATAL); 347} 348 349/* 350 * Set write options. 351 */ 352static int 353archive_compressor_xz_options(struct archive_write_filter *f, 354 const char *key, const char *value) 355{ 356 struct private_data *data = (struct private_data *)f->data; 357 358 if (strcmp(key, "compression-level") == 0) { 359 if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || 360 value[1] != '\0') 361 return (ARCHIVE_WARN); 362 data->compression_level = value[0] - '0'; 363 if (data->compression_level > 6) 364 data->compression_level = 6; 365 return (ARCHIVE_OK); 366 } 367 368 return (ARCHIVE_WARN); 369} 370 371/* 372 * Write data to the compressed stream. 373 */ 374static int 375archive_compressor_xz_write(struct archive_write_filter *f, 376 const void *buff, size_t length) 377{ 378 struct private_data *data = (struct private_data *)f->data; 379 int ret; 380 381 /* Update statistics */ 382 data->total_in += length; 383 if (f->code == ARCHIVE_COMPRESSION_LZIP) 384 data->crc32 = lzma_crc32(buff, length, data->crc32); 385 386 /* Compress input data to output buffer */ 387 data->stream.next_in = buff; 388 data->stream.avail_in = length; 389 if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK) 390 return (ret); 391 392 return (ARCHIVE_OK); 393} 394 395 396/* 397 * Finish the compression... 398 */ 399static int 400archive_compressor_xz_close(struct archive_write_filter *f) 401{ 402 struct private_data *data = (struct private_data *)f->data; 403 int ret, r1; 404 405 ret = drive_compressor(f, data, 1); 406 if (ret == ARCHIVE_OK) { 407 data->total_out += 408 data->compressed_buffer_size - data->stream.avail_out; 409 ret = __archive_write_filter(f->next_filter, 410 data->compressed, 411 data->compressed_buffer_size - data->stream.avail_out); 412 if (f->code == ARCHIVE_COMPRESSION_LZIP && ret == ARCHIVE_OK) { 413 archive_le32enc(data->compressed, data->crc32); 414 archive_le64enc(data->compressed+4, data->total_in); 415 archive_le64enc(data->compressed+12, data->total_out + 20); 416 ret = __archive_write_filter(f->next_filter, 417 data->compressed, 20); 418 } 419 } 420 lzma_end(&(data->stream)); 421 r1 = __archive_write_close_filter(f->next_filter); 422 return (r1 < ret ? r1 : ret); 423} 424 425static int 426archive_compressor_xz_free(struct archive_write_filter *f) 427{ 428 struct private_data *data = (struct private_data *)f->data; 429 free(data->compressed); 430 free(data); 431 f->data = NULL; 432 return (ARCHIVE_OK); 433} 434 435/* 436 * Utility function to push input data through compressor, 437 * writing full output blocks as necessary. 438 * 439 * Note that this handles both the regular write case (finishing == 440 * false) and the end-of-archive case (finishing == true). 441 */ 442static int 443drive_compressor(struct archive_write_filter *f, 444 struct private_data *data, int finishing) 445{ 446 int ret; 447 448 for (;;) { 449 if (data->stream.avail_out == 0) { 450 data->total_out += data->compressed_buffer_size; 451 ret = __archive_write_filter(f->next_filter, 452 data->compressed, 453 data->compressed_buffer_size); 454 if (ret != ARCHIVE_OK) 455 return (ARCHIVE_FATAL); 456 data->stream.next_out = data->compressed; 457 data->stream.avail_out = data->compressed_buffer_size; 458 } 459 460 /* If there's nothing to do, we're done. */ 461 if (!finishing && data->stream.avail_in == 0) 462 return (ARCHIVE_OK); 463 464 ret = lzma_code(&(data->stream), 465 finishing ? LZMA_FINISH : LZMA_RUN ); 466 467 switch (ret) { 468 case LZMA_OK: 469 /* In non-finishing case, check if compressor 470 * consumed everything */ 471 if (!finishing && data->stream.avail_in == 0) 472 return (ARCHIVE_OK); 473 /* In finishing case, this return always means 474 * there's more work */ 475 break; 476 case LZMA_STREAM_END: 477 /* This return can only occur in finishing case. */ 478 if (finishing) 479 return (ARCHIVE_OK); 480 archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, 481 "lzma compression data error"); 482 return (ARCHIVE_FATAL); 483 case LZMA_MEMLIMIT_ERROR: 484 archive_set_error(f->archive, ENOMEM, 485 "lzma compression error: " 486 "%ju MiB would have been needed", 487 (uintmax_t)((lzma_memusage(&(data->stream)) 488 + 1024 * 1024 -1) 489 / (1024 * 1024))); 490 return (ARCHIVE_FATAL); 491 default: 492 /* Any other return value indicates an error. */ 493 archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, 494 "lzma compression failed:" 495 " lzma_code() call returned status %d", 496 ret); 497 return (ARCHIVE_FATAL); 498 } 499 } 500} 501 502#endif /* HAVE_LZMA_H */ 503