src/liblzma/common/stream_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       stream_decoder.c
   4 /// \brief      Decodes .lzma Streams
   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 "stream_decoder.h"
  21 #include "block_decoder.h"
  22
  23
  24 struct lzma_coder_s {
  25         enum {
  26                 SEQ_STREAM_HEADER,
  27                 SEQ_BLOCK_HEADER,
  28                 SEQ_BLOCK,
  29                 SEQ_INDEX,
  30                 SEQ_STREAM_FOOTER,
  31                 SEQ_STREAM_PADDING,
  32         } sequence;
  33
  34         /// Block or Metadata decoder. This takes little memory and the same
  35         /// data structure can be used to decode every Block Header, so it's
  36         /// a good idea to have a separate lzma_next_coder structure for it.
  37         lzma_next_coder block_decoder;
  38
  39         /// Block options decoded by the Block Header decoder and used by
  40         /// the Block decoder.
  41         lzma_block block_options;
  42
  43         /// Stream Flags from Stream Header
  44         lzma_stream_flags stream_flags;
  45
  46         /// Index is hashed so that it can be compared to the sizes of Blocks
  47         /// with O(1) memory usage.
  48         lzma_index_hash *index_hash;
  49
  50         /// Memory usage limit
  51         uint64_t memlimit;
  52
  53         /// If true, LZMA_NO_CHECK is returned if the Stream has
  54         /// no integrity check.
  55         bool tell_no_check;
  56
  57         /// If true, LZMA_UNSUPPORTED_CHECK is returned if the Stream has
  58         /// an integrity check that isn't supported by this liblzma build.
  59         bool tell_unsupported_check;
  60
  61         /// If true, LZMA_GET_CHECK is returned after decoding Stream Header.
  62         bool tell_any_check;
  63
  64         /// If true, we will decode concatenated Streams that possibly have
  65         /// Stream Padding between or after them. LZMA_STREAM_END is returned
  66         /// once the application isn't giving us any new input, and we aren't
  67         /// in the middle of a Stream, and possible Stream Padding is a
  68         /// multiple of four bytes.
  69         bool concatenated;
  70
  71         /// When decoding concatenated Streams, this is true as long as we
  72         /// are decoding the first Stream. This is needed to avoid misleading
  73         /// LZMA_FORMAT_ERROR in case the later Streams don't have valid magic
  74         /// bytes.
  75         bool first_stream;
  76
  77         /// Write position in buffer[] and position in Stream Padding
  78         size_t pos;
  79
  80         /// Buffer to hold Stream Header, Block Header, and Stream Footer.
  81         /// Block Header has biggest maximum size.
  82         uint8_t buffer[LZMA_BLOCK_HEADER_SIZE_MAX];
  83 };
  84
  85
  86 static lzma_ret
  87 stream_decoder_reset(lzma_coder *coder, lzma_allocator *allocator)
  88 {
  89         // Initialize the Index hash used to verify the Index.
  90         coder->index_hash = lzma_index_hash_init(coder->index_hash, allocator);
  91         if (coder->index_hash == NULL)
  92                 return LZMA_MEM_ERROR;
  93
  94         // Reset the rest of the variables.
  95         coder->sequence = SEQ_STREAM_HEADER;
  96         coder->block_options.filters = NULL;
  97         coder->pos = 0;
  98
  99         return LZMA_OK;
 100 }
 101
 102
 103 static lzma_ret
 104 stream_decode(lzma_coder *coder, lzma_allocator *allocator,
 105                 const uint8_t *restrict in, size_t *restrict in_pos,
 106                 size_t in_size, uint8_t *restrict out,
 107                 size_t *restrict out_pos, size_t out_size, lzma_action action)
 108 {
 109         // When decoding the actual Block, it may be able to produce more
 110         // output even if we don't give it any new input.
 111         while (true)
 112         switch (coder->sequence) {
 113         case SEQ_STREAM_HEADER: {
 114                 // Copy the Stream Header to the internal buffer.
 115                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 116                                 LZMA_STREAM_HEADER_SIZE);
 117
 118                 // Return if we didn't get the whole Stream Header yet.
 119                 if (coder->pos < LZMA_STREAM_HEADER_SIZE)
 120                         return LZMA_OK;
 121
 122                 coder->pos = 0;
 123
 124                 // Decode the Stream Header.
 125                 const lzma_ret ret = lzma_stream_header_decode(
 126                                 &coder->stream_flags, coder->buffer);
 127                 if (ret != LZMA_OK)
 128                         return ret == LZMA_FORMAT_ERROR && !coder->first_stream
 129                                         ? LZMA_DATA_ERROR : ret;
 130
 131                 // Copy the type of the Check so that Block Header and Block
 132                 // decoders see it.
 133                 coder->block_options.check = coder->stream_flags.check;
 134
 135                 // Even if we return LZMA_*_CHECK below, we want
 136                 // to continue from Block Header decoding.
 137                 coder->sequence = SEQ_BLOCK_HEADER;
 138
 139                 // Detect if there's no integrity check or if it is
 140                 // unsupported if those were requested by the application.
 141                 if (coder->tell_no_check && coder->stream_flags.check
 142                                 == LZMA_CHECK_NONE)
 143                         return LZMA_NO_CHECK;
 144
 145                 if (coder->tell_unsupported_check
 146                                 && !lzma_check_is_supported(
 147                                         coder->stream_flags.check))
 148                         return LZMA_UNSUPPORTED_CHECK;
 149
 150                 if (coder->tell_any_check)
 151                         return LZMA_GET_CHECK;
 152         }
 153
 154         // Fall through
 155
 156         case SEQ_BLOCK_HEADER: {
 157                 if (*in_pos >= in_size)
 158                         return LZMA_OK;
 159
 160                 if (coder->pos == 0) {
 161                         // Detect if it's Index.
 162                         if (in[*in_pos] == 0x00) {
 163                                 coder->sequence = SEQ_INDEX;
 164                                 break;
 165                         }
 166
 167                         // Calculate the size of the Block Header. Note that
 168                         // Block Header decoder wants to see this byte too
 169                         // so don't advance *in_pos.
 170                         coder->block_options.header_size
 171                                         = lzma_block_header_size_decode(
 172                                                 in[*in_pos]);
 173                 }
 174
 175                 // Copy the Block Header to the internal buffer.
 176                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 177                                 coder->block_options.header_size);
 178
 179                 // Return if we didn't get the whole Block Header yet.
 180                 if (coder->pos < coder->block_options.header_size)
 181                         return LZMA_OK;
 182
 183                 coder->pos = 0;
 184
 185                 // Set up a buffer to hold the filter chain. Block Header
 186                 // decoder will initialize all members of this array so
 187                 // we don't need to do it here.
 188                 lzma_filter filters[LZMA_BLOCK_FILTERS_MAX + 1];
 189                 coder->block_options.filters = filters;
 190
 191                 // Decode the Block Header.
 192                 return_if_error(lzma_block_header_decode(&coder->block_options,
 193                                 allocator, coder->buffer));
 194
 195                 // Check the memory usage limit.
 196                 const uint64_t memusage = lzma_memusage_decoder(filters);
 197                 lzma_ret ret;
 198
 199                 if (memusage == UINT64_MAX) {
 200                         // One or more unknown Filter IDs.
 201                         ret = LZMA_HEADER_ERROR;
 202                 } else if (memusage > coder->memlimit) {
 203                         // The chain would need too much memory.
 204                         ret = LZMA_MEMLIMIT_ERROR;
 205                 } else {
 206                         // Memory usage is OK. Initialize the Block decoder.
 207                         ret = lzma_block_decoder_init(
 208                                         &coder->block_decoder,
 209                                         allocator, &coder->block_options);
 210                 }
 211
 212                 // Free the allocated filter options since they are needed
 213                 // only to initialize the Block decoder.
 214                 for (size_t i = 0; i < LZMA_BLOCK_FILTERS_MAX; ++i)
 215                         lzma_free(filters[i].options, allocator);
 216
 217                 coder->block_options.filters = NULL;
 218
 219                 // Check if memory usage calculation and Block enocoder
 220                 // initialization succeeded.
 221                 if (ret != LZMA_OK)
 222                         return ret;
 223
 224                 coder->sequence = SEQ_BLOCK;
 225         }
 226
 227         // Fall through
 228
 229         case SEQ_BLOCK: {
 230                 const lzma_ret ret = coder->block_decoder.code(
 231                                 coder->block_decoder.coder, allocator,
 232                                 in, in_pos, in_size, out, out_pos, out_size,
 233                                 action);
 234
 235                 if (ret != LZMA_STREAM_END)
 236                         return ret;
 237
 238                 // Block decoded successfully. Add the new size pair to
 239                 // the Index hash.
 240                 return_if_error(lzma_index_hash_append(coder->index_hash,
 241                                 lzma_block_total_size_get(
 242                                         &coder->block_options),
 243                                 coder->block_options.uncompressed_size));
 244
 245                 coder->sequence = SEQ_BLOCK_HEADER;
 246                 break;
 247         }
 248
 249         case SEQ_INDEX: {
 250                 // If we don't have any input, don't call
 251                 // lzma_index_hash_decode() since it would return
 252                 // LZMA_BUF_ERROR, which we must not do here.
 253                 if (*in_pos >= in_size)
 254                         return LZMA_OK;
 255
 256                 // Decode the Index and compare it to the hash calculated
 257                 // from the sizes of the Blocks (if any).
 258                 const lzma_ret ret = lzma_index_hash_decode(coder->index_hash,
 259                                 in, in_pos, in_size);
 260                 if (ret != LZMA_STREAM_END)
 261                         return ret;
 262
 263                 coder->sequence = SEQ_STREAM_FOOTER;
 264         }
 265
 266         // Fall through
 267
 268         case SEQ_STREAM_FOOTER:
 269                 // Copy the Stream Footer to the internal buffer.
 270                 lzma_bufcpy(in, in_pos, in_size, coder->buffer, &coder->pos,
 271                                 LZMA_STREAM_HEADER_SIZE);
 272
 273                 // Return if we didn't get the whole Stream Footer yet.
 274                 if (coder->pos < LZMA_STREAM_HEADER_SIZE)
 275                         return LZMA_OK;
 276
 277                 coder->pos = 0;
 278
 279                 // Decode the Stream Footer. The decoder gives
 280                 // LZMA_FORMAT_ERROR if the magic bytes don't match,
 281                 // so convert that return code to LZMA_DATA_ERROR.
 282                 lzma_stream_flags footer_flags;
 283                 const lzma_ret ret = lzma_stream_footer_decode(
 284                                 &footer_flags, coder->buffer);
 285                 if (ret != LZMA_OK)
 286                         return ret == LZMA_FORMAT_ERROR
 287                                         ? LZMA_DATA_ERROR : ret;
 288
 289                 // Check that Index Size stored in the Stream Footer matches
 290                 // the real size of the Index field.
 291                 if (lzma_index_hash_size(coder->index_hash)
 292                                 != footer_flags.backward_size)
 293                         return LZMA_DATA_ERROR;
 294
 295                 // Compare that the Stream Flags fields are identical in
 296                 // both Stream Header and Stream Footer.
 297                 return_if_error(lzma_stream_flags_compare(
 298                                 &coder->stream_flags, &footer_flags));
 299
 300                 if (!coder->concatenated)
 301                         return LZMA_STREAM_END;
 302
 303                 coder->sequence = SEQ_STREAM_PADDING;
 304
 305         // Fall through
 306
 307         case SEQ_STREAM_PADDING:
 308                 assert(coder->concatenated);
 309
 310                 // Skip over possible Stream Padding.
 311                 while (true) {
 312                         if (*in_pos >= in_size) {
 313                                 // Unless LZMA_FINISH was used, we cannot
 314                                 // know if there's more input coming later.
 315                                 if (action != LZMA_FINISH)
 316                                         return LZMA_OK;
 317
 318                                 // Stream Padding must be a multiple of
 319                                 // four bytes.
 320                                 return coder->pos == 0
 321                                                 ? LZMA_STREAM_END
 322                                                 : LZMA_DATA_ERROR;
 323                         }
 324
 325                         // If the byte is not zero, it probably indicates
 326                         // beginning of a new Stream (or the file is corrupt).
 327                         if (in[*in_pos] != 0x00)
 328                                 break;
 329
 330                         ++*in_pos;
 331                         coder->pos = (coder->pos + 1) & 3;
 332                 }
 333
 334                 // Stream Padding must be a multiple of four bytes (empty
 335                 // Stream Padding is OK).
 336                 if (coder->pos != 0) {
 337                         ++*in_pos;
 338                         return LZMA_DATA_ERROR;
 339                 }
 340
 341                 // Prepare to decode the next Stream.
 342                 return_if_error(stream_decoder_reset(coder, allocator));
 343                 break;
 344
 345         default:
 346                 assert(0);
 347                 return LZMA_PROG_ERROR;
 348         }
 349
 350         return LZMA_OK;
 351 }
 352
 353
 354 static void
 355 stream_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 356 {
 357         lzma_next_end(&coder->block_decoder, allocator);
 358         lzma_index_hash_end(coder->index_hash, allocator);
 359         lzma_free(coder, allocator);
 360         return;
 361 }
 362
 363
 364 static lzma_check
 365 stream_decoder_get_check(const lzma_coder *coder)
 366 {
 367         return coder->stream_flags.check;
 368 }
 369
 370
 371 extern lzma_ret
 372 lzma_stream_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 373                 uint64_t memlimit, uint32_t flags)
 374 {
 375         lzma_next_coder_init(lzma_stream_decoder_init, next, allocator);
 376
 377         if (flags & ~LZMA_SUPPORTED_FLAGS)
 378                 return LZMA_HEADER_ERROR;
 379
 380         if (next->coder == NULL) {
 381                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 382                 if (next->coder == NULL)
 383                         return LZMA_MEM_ERROR;
 384
 385                 next->code = &stream_decode;
 386                 next->end = &stream_decoder_end;
 387                 next->get_check = &stream_decoder_get_check;
 388
 389                 next->coder->block_decoder = LZMA_NEXT_CODER_INIT;
 390                 next->coder->index_hash = NULL;
 391         }
 392
 393         next->coder->memlimit = memlimit;
 394         next->coder->tell_no_check = (flags & LZMA_TELL_NO_CHECK) != 0;
 395         next->coder->tell_unsupported_check
 396                         = (flags & LZMA_TELL_UNSUPPORTED_CHECK) != 0;
 397         next->coder->tell_any_check = (flags & LZMA_TELL_ANY_CHECK) != 0;
 398         next->coder->concatenated = (flags & LZMA_CONCATENATED) != 0;
 399
 400         return stream_decoder_reset(next->coder, allocator);
 401 }
 402
 403
 404 extern LZMA_API lzma_ret
 405 lzma_stream_decoder(lzma_stream *strm, uint64_t memlimit, uint32_t flags)
 406 {
 407         lzma_next_strm_init(lzma_stream_decoder_init, strm, memlimit, flags);
 408
 409         strm->internal->supported_actions[LZMA_RUN] = true;
 410         strm->internal->supported_actions[LZMA_FINISH] = true;
 411
 412         return LZMA_OK;
 413 }