src/liblzma/common/metadata_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       metadata_decoder.c
   4 /// \brief      Decodes metadata stored in Metadata Blocks
   5 //
   6 //  Copyright (C) 2007 Lasse Collin
   7 //
   8 //  This library is free software; you can redistribute it and/or
   9 //  modify it under the terms of the GNU Lesser General Public
  10 //  License as published by the Free Software Foundation; either
  11 //  version 2.1 of the License, or (at your option) any later version.
  12 //
  13 //  This library is distributed in the hope that it will be useful,
  14 //  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 //  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16 //  Lesser General Public License for more details.
  17 //
  18 ///////////////////////////////////////////////////////////////////////////////
  19
  20 #include "metadata_decoder.h"
  21 #include "block_decoder.h"
  22
  23
  24 /// Maximum size of a single Extra Record. Again, this is mostly to make
  25 /// sure that the parsed lzma_vli fits into size_t. Still, maybe this should
  26 /// be smaller.
  27 #define EXTRA_SIZE_MAX (SIZE_MAX / 4)
  28
  29
  30 struct lzma_coder_s {
  31         enum {
  32                 SEQ_FLAGS,
  33                 SEQ_HEADER_METADATA_SIZE,
  34                 SEQ_TOTAL_SIZE,
  35                 SEQ_UNCOMPRESSED_SIZE,
  36                 SEQ_INDEX_COUNT,
  37                 SEQ_INDEX_ALLOC,
  38                 SEQ_INDEX_TOTAL_SIZE,
  39                 SEQ_INDEX_UNCOMPRESSED_SIZE,
  40                 SEQ_EXTRA_PREPARE,
  41                 SEQ_EXTRA_ALLOC,
  42                 SEQ_EXTRA_ID,
  43                 SEQ_EXTRA_SIZE,
  44                 SEQ_EXTRA_DATA_ALLOC,
  45                 SEQ_EXTRA_DATA_COPY,
  46                 SEQ_EXTRA_DUMMY_ALLOC,
  47                 SEQ_EXTRA_DUMMY_ID,
  48                 SEQ_EXTRA_DUMMY_SIZE,
  49                 SEQ_EXTRA_DUMMY_COPY,
  50         } sequence;
  51
  52         /// Number of "things" left to be parsed. If we hit end of input
  53         /// when this isn't zero, we have corrupt Metadata Block.
  54         size_t todo_count;
  55
  56         /// Position in variable-length integers
  57         size_t pos;
  58
  59         /// Temporary variable needed to decode variables whose type
  60         /// is size_t instead of lzma_vli.
  61         lzma_vli tmp;
  62
  63         /// Pointer to target structure to hold the parsed results.
  64         lzma_metadata *metadata;
  65
  66         /// The Index Record we currently are parsing
  67         lzma_index *index_current;
  68
  69         /// Number of Records in Index
  70         size_t index_count;
  71
  72         /// Sum of Total Size fields in the Index
  73         lzma_vli index_total_size;
  74
  75         /// Sum of Uncompressed Size fields in the Index
  76         lzma_vli index_uncompressed_size;
  77
  78         /// True if Extra is present.
  79         bool has_extra;
  80
  81         /// True if we have been requested to store the Extra to *metadata.
  82         bool want_extra;
  83
  84         /// Pointer to the end of the Extra Record list.
  85         lzma_extra *extra_tail;
  86
  87         /// Dummy Extra Record used when only verifying integrity of Extra
  88         /// (not storing it to RAM).
  89         lzma_extra extra_dummy;
  90
  91         /// Block decoder
  92         lzma_next_coder block_decoder;
  93
  94         /// buffer[buffer_pos] is the next byte to process.
  95         size_t buffer_pos;
  96
  97         /// buffer[buffer_size] is the first byte to not process.
  98         size_t buffer_size;
  99
 100         /// Temporary buffer to which encoded Metadata is read before
 101         /// it is parsed.
 102         uint8_t buffer[LZMA_BUFFER_SIZE];
 103 };
 104
 105
 106 /// Reads a variable-length integer to coder->num.
 107 #define read_vli(num) \
 108 do { \
 109         const lzma_ret ret = lzma_vli_decode( \
 110                         &num, &coder->pos, \
 111                         coder->buffer, &coder->buffer_pos, \
 112                         coder->buffer_size); \
 113         if (ret != LZMA_STREAM_END) \
 114                 return ret; \
 115         \
 116         coder->pos = 0; \
 117 } while (0)
 118
 119
 120 static lzma_ret
 121 process(lzma_coder *coder, lzma_allocator *allocator)
 122 {
 123         while (coder->buffer_pos < coder->buffer_size)
 124         switch (coder->sequence) {
 125         case SEQ_FLAGS:
 126                 // Reserved bits must be unset.
 127                 if (coder->buffer[coder->buffer_pos] & 0x70)
 128                         return LZMA_HEADER_ERROR;
 129
 130                 // If Size of Header Metadata is present, prepare the
 131                 // variable for variable-length integer decoding. Otherwise
 132                 // set it to LZMA_VLI_VALUE_UNKNOWN to indicate that the
 133                 // field isn't present.
 134                 if (coder->buffer[coder->buffer_pos] & 0x01) {
 135                         coder->metadata->header_metadata_size = 0;
 136                         ++coder->todo_count;
 137                 }
 138
 139                 if (coder->buffer[coder->buffer_pos] & 0x02) {
 140                         coder->metadata->total_size = 0;
 141                         ++coder->todo_count;
 142                 }
 143
 144                 if (coder->buffer[coder->buffer_pos] & 0x04) {
 145                         coder->metadata->uncompressed_size = 0;
 146                         ++coder->todo_count;
 147                 }
 148
 149                 if (coder->buffer[coder->buffer_pos] & 0x08) {
 150                         // Setting index_count to 1 is just to indicate that
 151                         // Index is present. The real size is parsed later.
 152                         coder->index_count = 1;
 153                         ++coder->todo_count;
 154                 }
 155
 156                 coder->has_extra = (coder->buffer[coder->buffer_pos] & 0x80)
 157                                 != 0;
 158
 159                 ++coder->buffer_pos;
 160                 coder->sequence = SEQ_HEADER_METADATA_SIZE;
 161                 break;
 162
 163         case SEQ_HEADER_METADATA_SIZE:
 164                 if (coder->metadata->header_metadata_size
 165                                 != LZMA_VLI_VALUE_UNKNOWN) {
 166                         read_vli(coder->metadata->header_metadata_size);
 167
 168                         if (coder->metadata->header_metadata_size == 0)
 169                                 return LZMA_DATA_ERROR;
 170
 171                         --coder->todo_count;
 172                 }
 173
 174                 coder->sequence = SEQ_TOTAL_SIZE;
 175                 break;
 176
 177         case SEQ_TOTAL_SIZE:
 178                 if (coder->metadata->total_size != LZMA_VLI_VALUE_UNKNOWN) {
 179                         read_vli(coder->metadata->total_size);
 180
 181                         if (coder->metadata->total_size == 0)
 182                                 return LZMA_DATA_ERROR;
 183
 184                         --coder->todo_count;
 185                 }
 186
 187                 coder->sequence = SEQ_UNCOMPRESSED_SIZE;
 188                 break;
 189
 190         case SEQ_UNCOMPRESSED_SIZE:
 191                 if (coder->metadata->uncompressed_size
 192                                 != LZMA_VLI_VALUE_UNKNOWN) {
 193                         read_vli(coder->metadata->uncompressed_size);
 194                         --coder->todo_count;
 195                 }
 196
 197                 coder->sequence = SEQ_INDEX_COUNT;
 198                 break;
 199
 200         case SEQ_INDEX_COUNT:
 201                 if (coder->index_count == 0) {
 202                         coder->sequence = SEQ_EXTRA_PREPARE;
 203                         break;
 204                 }
 205
 206                 read_vli(coder->tmp);
 207
 208                 // Index must not be empty nor far too big (wouldn't fit
 209                 // in RAM).
 210                 if (coder->tmp == 0 || coder->tmp
 211                                 >= SIZE_MAX / sizeof(lzma_index))
 212                         return LZMA_DATA_ERROR;
 213
 214                 coder->index_count = (size_t)(coder->tmp);
 215                 coder->tmp = 0;
 216
 217                 coder->sequence = SEQ_INDEX_ALLOC;
 218                 break;
 219
 220         case SEQ_INDEX_ALLOC: {
 221                 lzma_index *i = lzma_alloc(sizeof(lzma_index), allocator);
 222                 if (i == NULL)
 223                         return LZMA_MEM_ERROR;
 224
 225                 i->total_size = 0;
 226                 i->uncompressed_size = 0;
 227                 i->next = NULL;
 228
 229                 if (coder->metadata->index == NULL)
 230                         coder->metadata->index = i;
 231                 else
 232                         coder->index_current->next = i;
 233
 234                 coder->index_current = i;
 235
 236                 coder->sequence = SEQ_INDEX_TOTAL_SIZE;
 237         }
 238
 239         // Fall through
 240
 241         case SEQ_INDEX_TOTAL_SIZE: {
 242                 read_vli(coder->index_current->total_size);
 243
 244                 coder->index_total_size += coder->index_current->total_size;
 245                 if (coder->index_total_size > LZMA_VLI_VALUE_MAX)
 246                         return LZMA_DATA_ERROR;
 247
 248                 // No Block can have Total Size of zero bytes.
 249                 if (coder->index_current->total_size == 0)
 250                         return LZMA_DATA_ERROR;
 251
 252                 if (--coder->index_count == 0) {
 253                         // If Total Size is present, it must match the sum
 254                         // of Total Sizes in Index.
 255                         if (coder->metadata->total_size
 256                                                 != LZMA_VLI_VALUE_UNKNOWN
 257                                         && coder->metadata->total_size
 258                                                 != coder->index_total_size)
 259                                 return LZMA_DATA_ERROR;
 260
 261                         coder->index_current = coder->metadata->index;
 262                         coder->sequence = SEQ_INDEX_UNCOMPRESSED_SIZE;
 263                 } else {
 264                         coder->sequence = SEQ_INDEX_ALLOC;
 265                 }
 266
 267                 break;
 268         }
 269
 270         case SEQ_INDEX_UNCOMPRESSED_SIZE: {
 271                 read_vli(coder->index_current->uncompressed_size);
 272
 273                 coder->index_uncompressed_size
 274                                 += coder->index_current->uncompressed_size;
 275                 if (coder->index_uncompressed_size > LZMA_VLI_VALUE_MAX)
 276                         return LZMA_DATA_ERROR;
 277
 278                 coder->index_current = coder->index_current->next;
 279                 if (coder->index_current == NULL) {
 280                         if (coder->metadata->uncompressed_size
 281                                                 != LZMA_VLI_VALUE_UNKNOWN
 282                                         && coder->metadata->uncompressed_size
 283                                         != coder->index_uncompressed_size)
 284                                 return LZMA_DATA_ERROR;
 285
 286                         --coder->todo_count;
 287                         coder->sequence = SEQ_EXTRA_PREPARE;
 288                 }
 289
 290                 break;
 291         }
 292
 293         case SEQ_EXTRA_PREPARE:
 294                 assert(coder->todo_count == 0);
 295
 296                 // If we get here, we have at least one byte of input left.
 297                 // If "Extra is present" flag is unset in Metadata Flags,
 298                 // it means that there is some garbage and we return an error.
 299                 if (!coder->has_extra)
 300                         return LZMA_DATA_ERROR;
 301
 302                 if (!coder->want_extra) {
 303                         coder->extra_tail = &coder->extra_dummy;
 304                         coder->sequence = SEQ_EXTRA_DUMMY_ALLOC;
 305                         break;
 306                 }
 307
 308                 coder->sequence = SEQ_EXTRA_ALLOC;
 309
 310         // Fall through
 311
 312         case SEQ_EXTRA_ALLOC: {
 313                 lzma_extra *e = lzma_alloc(sizeof(lzma_extra), allocator);
 314                 if (e == NULL)
 315                         return LZMA_MEM_ERROR;
 316
 317                 e->next = NULL;
 318                 e->id = 0;
 319                 e->size = 0;
 320                 e->data = NULL;
 321
 322                 if (coder->metadata->extra == NULL)
 323                         coder->metadata->extra = e;
 324                 else
 325                         coder->extra_tail->next = e;
 326
 327                 coder->extra_tail = e;
 328
 329                 coder->todo_count = 1;
 330                 coder->sequence = SEQ_EXTRA_ID;
 331         }
 332
 333         // Fall through
 334
 335         case SEQ_EXTRA_ID:
 336         case SEQ_EXTRA_DUMMY_ID:
 337                 read_vli(coder->extra_tail->id);
 338
 339                 if (coder->extra_tail->id == 0) {
 340                         coder->extra_tail->size = 0;
 341                         coder->extra_tail->data = NULL;
 342                         coder->todo_count = 0;
 343                         --coder->sequence;
 344                 } else {
 345                         ++coder->sequence;
 346                 }
 347
 348                 break;
 349
 350         case SEQ_EXTRA_SIZE:
 351         case SEQ_EXTRA_DUMMY_SIZE:
 352                 read_vli(coder->tmp);
 353
 354                 if (coder->tmp == 0) {
 355                         // We have no Data in the Extra Record. Don't
 356                         // allocate any memory for it. Go back to
 357                         // SEQ_EXTRA_ALLOC or SEQ_EXTRA_DUMMY_ALLOC.
 358                         coder->tmp = 0;
 359                         coder->sequence -= 2;
 360                         coder->todo_count = 0;
 361                 } else {
 362                         ++coder->sequence;
 363                 }
 364
 365                 break;
 366
 367         case SEQ_EXTRA_DATA_ALLOC: {
 368                 if (coder->tmp > EXTRA_SIZE_MAX)
 369                         return LZMA_DATA_ERROR;
 370
 371                 coder->extra_tail->size = (size_t)(coder->tmp);
 372                 coder->tmp = 0;
 373
 374                 // We reserve space for the trailing '\0' too.
 375                 uint8_t *d = lzma_alloc((size_t)(coder->extra_tail->size) + 1,
 376                                 allocator);
 377                 if (d == NULL)
 378                         return LZMA_MEM_ERROR;
 379
 380                 coder->extra_tail->data = d;
 381                 coder->sequence = SEQ_EXTRA_DATA_COPY;
 382         }
 383
 384         // Fall through
 385
 386         case SEQ_EXTRA_DATA_COPY:
 387                 bufcpy(coder->buffer, &coder->buffer_pos, coder->buffer_size,
 388                                 coder->extra_tail->data, &coder->pos,
 389                                 (size_t)(coder->extra_tail->size));
 390
 391                 if ((size_t)(coder->extra_tail->size) == coder->pos) {
 392                         coder->extra_tail->data[coder->pos] = '\0';
 393                         coder->pos = 0;
 394                         coder->todo_count = 0;
 395                         coder->sequence = SEQ_EXTRA_ALLOC;
 396                 }
 397
 398                 break;
 399
 400         case SEQ_EXTRA_DUMMY_ALLOC:
 401                 // Not really alloc, just initialize the dummy entry.
 402                 coder->extra_dummy = (lzma_extra){
 403                         .next = NULL,
 404                         .id = 0,
 405                         .size = 0,
 406                         .data = NULL,
 407                 };
 408
 409                 coder->todo_count = 1;
 410                 coder->sequence = SEQ_EXTRA_DUMMY_ID;
 411                 break;
 412
 413         case SEQ_EXTRA_DUMMY_COPY: {
 414                 // Simply skip as many bytes as indicated by Extra Record Size.
 415                 // We don't check lzma_extra_size_max because we don't
 416                 // allocate any memory to hold the data.
 417                 const size_t in_avail = coder->buffer_size - coder->buffer_pos;
 418                 const size_t skip = MIN((lzma_vli)(in_avail), coder->tmp);
 419                 coder->buffer_pos += skip;
 420                 coder->tmp -= skip;
 421
 422                 if (coder->tmp == 0) {
 423                         coder->todo_count = 0;
 424                         coder->sequence = SEQ_EXTRA_DUMMY_ALLOC;
 425                 }
 426
 427                 break;
 428         }
 429
 430         default:
 431                 return LZMA_PROG_ERROR;
 432         }
 433
 434         return LZMA_OK;
 435 }
 436
 437
 438 static lzma_ret
 439 metadata_decode(lzma_coder *coder, lzma_allocator *allocator,
 440                 const uint8_t *restrict in, size_t *restrict in_pos,
 441                 size_t in_size, uint8_t *restrict out lzma_attribute((unused)),
 442                 size_t *restrict out_pos lzma_attribute((unused)),
 443                 size_t out_size lzma_attribute((unused)),
 444                 lzma_action action lzma_attribute((unused)))
 445 {
 446         bool end_was_reached = false;
 447
 448         while (true) {
 449                 // Fill the buffer if it is empty.
 450                 if (coder->buffer_pos == coder->buffer_size) {
 451                         coder->buffer_pos = 0;
 452                         coder->buffer_size = 0;
 453
 454                         const lzma_ret ret = coder->block_decoder.code(
 455                                         coder->block_decoder.coder, allocator,
 456                                         in, in_pos, in_size, coder->buffer,
 457                                         &coder->buffer_size, LZMA_BUFFER_SIZE,
 458                                         LZMA_RUN);
 459
 460                         switch (ret) {
 461                         case LZMA_OK:
 462                                 // Return immediatelly if we got no new data.
 463                                 if (coder->buffer_size == 0)
 464                                         return LZMA_OK;
 465
 466                                 break;
 467
 468                         case LZMA_STREAM_END:
 469                                 end_was_reached = true;
 470                                 break;
 471
 472                         default:
 473                                 return ret;
 474                         }
 475                 }
 476
 477                 // Process coder->buffer.
 478                 const lzma_ret ret = process(coder, allocator);
 479                 if (ret != LZMA_OK)
 480                         return ret;
 481
 482                 // On success, process() eats all the input.
 483                 assert(coder->buffer_pos == coder->buffer_size);
 484
 485                 if (end_was_reached) {
 486                         // Check that the sequence is not in the
 487                         // middle of anything.
 488                         if (coder->todo_count != 0)
 489                                 return LZMA_DATA_ERROR;
 490
 491                         // If Size of Header Metadata Block was not
 492                         // present, we use zero as its size instead
 493                         // of LZMA_VLI_VALUE_UNKNOWN.
 494                         if (coder->metadata->header_metadata_size
 495                                         == LZMA_VLI_VALUE_UNKNOWN)
 496                                 coder->metadata->header_metadata_size = 0;
 497
 498                         return LZMA_STREAM_END;
 499                 }
 500         }
 501 }
 502
 503
 504 static void
 505 metadata_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 506 {
 507         lzma_next_coder_end(&coder->block_decoder, allocator);
 508         lzma_free(coder, allocator);
 509         return;
 510 }
 511
 512
 513 static lzma_ret
 514 metadata_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 515                 lzma_options_block *options, lzma_metadata *metadata,
 516                 bool want_extra)
 517 {
 518         if (options == NULL || metadata == NULL)
 519                 return LZMA_PROG_ERROR;
 520
 521         if (next->coder == NULL) {
 522                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 523                 if (next->coder == NULL)
 524                         return LZMA_MEM_ERROR;
 525
 526                 next->code = &metadata_decode;
 527                 next->end = &metadata_decoder_end;
 528                 next->coder->block_decoder = LZMA_NEXT_CODER_INIT;
 529         }
 530
 531         metadata->header_metadata_size = LZMA_VLI_VALUE_UNKNOWN;
 532         metadata->total_size = LZMA_VLI_VALUE_UNKNOWN;
 533         metadata->uncompressed_size = LZMA_VLI_VALUE_UNKNOWN;
 534         metadata->index = NULL;
 535         metadata->extra = NULL;
 536
 537         next->coder->sequence = SEQ_FLAGS;
 538         next->coder->todo_count = 0;
 539         next->coder->pos = 0;
 540         next->coder->tmp = 0;
 541         next->coder->metadata = metadata;
 542         next->coder->index_current = NULL;
 543         next->coder->index_count = 0;
 544         next->coder->index_total_size = 0;
 545         next->coder->index_uncompressed_size = 0;
 546         next->coder->want_extra = want_extra;
 547         next->coder->extra_tail = NULL;
 548         next->coder->buffer_pos = 0;
 549         next->coder->buffer_size = 0;
 550
 551         return lzma_block_decoder_init(
 552                         &next->coder->block_decoder, allocator, options);
 553 }
 554
 555
 556 extern lzma_ret
 557 lzma_metadata_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 558                 lzma_options_block *options, lzma_metadata *metadata,
 559                 bool want_extra)
 560 {
 561         lzma_next_coder_init(metadata_decoder_init, next, allocator,
 562                         options, metadata, want_extra);
 563 }
 564
 565
 566 extern LZMA_API lzma_ret
 567 lzma_metadata_decoder(lzma_stream *strm, lzma_options_block *options,
 568                 lzma_metadata *metadata, lzma_bool want_extra)
 569 {
 570         lzma_next_strm_init(strm, lzma_metadata_decoder_init,
 571                         options, metadata, want_extra);
 572
 573         strm->internal->supported_actions[LZMA_RUN] = true;
 574
 575         return LZMA_OK;
 576 }