src/liblzma/common/alone_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       alone_decoder.c
   4 /// \brief      Decoder for LZMA_Alone files
   5 //
   6 //  Author:     Lasse Collin
   7 //
   8 //  This file has been put into the public domain.
   9 //  You can do whatever you want with this file.
  10 //
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 #include "alone_decoder.h"
  14 #include "lzma_decoder.h"
  15 #include "lz_decoder.h"
  16
  17
  18 struct lzma_coder_s {
  19         lzma_next_coder next;
  20
  21         enum {
  22                 SEQ_PROPERTIES,
  23                 SEQ_DICTIONARY_SIZE,
  24                 SEQ_UNCOMPRESSED_SIZE,
  25                 SEQ_CODER_INIT,
  26                 SEQ_CODE,
  27         } sequence;
  28
  29         /// If true, reject files that are unlikely to be .lzma files.
  30         /// If false, more non-.lzma files get accepted and will give
  31         /// LZMA_DATA_ERROR either immediately or after a few output bytes.
  32         bool picky;
  33
  34         /// Position in the header fields
  35         size_t pos;
  36
  37         /// Uncompressed size decoded from the header
  38         lzma_vli uncompressed_size;
  39
  40         /// Memory usage limit
  41         uint64_t memlimit;
  42
  43         /// Amount of memory actually needed (only an estimate)
  44         uint64_t memusage;
  45
  46         /// Options decoded from the header needed to initialize
  47         /// the LZMA decoder
  48         lzma_options_lzma options;
  49 };
  50
  51
  52 static lzma_ret
  53 alone_decode(lzma_coder *coder,
  54                 lzma_allocator *allocator lzma_attribute((__unused__)),
  55                 const uint8_t *restrict in, size_t *restrict in_pos,
  56                 size_t in_size, uint8_t *restrict out,
  57                 size_t *restrict out_pos, size_t out_size,
  58                 lzma_action action)
  59 {
  60         while (*out_pos < out_size
  61                         && (coder->sequence == SEQ_CODE || *in_pos < in_size))
  62         switch (coder->sequence) {
  63         case SEQ_PROPERTIES:
  64                 if (lzma_lzma_lclppb_decode(&coder->options, in[*in_pos]))
  65                         return LZMA_FORMAT_ERROR;
  66
  67                 coder->sequence = SEQ_DICTIONARY_SIZE;
  68                 ++*in_pos;
  69                 break;
  70
  71         case SEQ_DICTIONARY_SIZE:
  72                 coder->options.dict_size
  73                                 |= (size_t)(in[*in_pos]) << (coder->pos * 8);
  74
  75                 if (++coder->pos == 4) {
  76                         if (coder->picky && coder->options.dict_size
  77                                         != UINT32_MAX) {
  78                                 // A hack to ditch tons of false positives:
  79                                 // We allow only dictionary sizes that are
  80                                 // 2^n or 2^n + 2^(n-1). LZMA_Alone created
  81                                 // only files with 2^n, but accepts any
  82                                 // dictionary size.
  83                                 uint32_t d = coder->options.dict_size - 1;
  84                                 d |= d >> 2;
  85                                 d |= d >> 3;
  86                                 d |= d >> 4;
  87                                 d |= d >> 8;
  88                                 d |= d >> 16;
  89                                 ++d;
  90
  91                                 if (d != coder->options.dict_size)
  92                                         return LZMA_FORMAT_ERROR;
  93                         }
  94
  95                         coder->pos = 0;
  96                         coder->sequence = SEQ_UNCOMPRESSED_SIZE;
  97                 }
  98
  99                 ++*in_pos;
 100                 break;
 101
 102         case SEQ_UNCOMPRESSED_SIZE:
 103                 coder->uncompressed_size
 104                                 |= (lzma_vli)(in[*in_pos]) << (coder->pos * 8);
 105                 ++*in_pos;
 106                 if (++coder->pos < 8)
 107                         break;
 108
 109                 // Another hack to ditch false positives: Assume that
 110                 // if the uncompressed size is known, it must be less
 111                 // than 256 GiB.
 112                 if (coder->picky
 113                                 && coder->uncompressed_size != LZMA_VLI_UNKNOWN
 114                                 && coder->uncompressed_size
 115                                         >= (LZMA_VLI_C(1) << 38))
 116                         return LZMA_FORMAT_ERROR;
 117
 118                 // Calculate the memory usage so that it is ready
 119                 // for SEQ_CODER_INIT.
 120                 coder->memusage = lzma_lzma_decoder_memusage(&coder->options)
 121                                 + LZMA_MEMUSAGE_BASE;
 122
 123                 coder->pos = 0;
 124                 coder->sequence = SEQ_CODER_INIT;
 125
 126         // Fall through
 127
 128         case SEQ_CODER_INIT: {
 129                 if (coder->memusage > coder->memlimit)
 130                         return LZMA_MEMLIMIT_ERROR;
 131
 132                 lzma_filter_info filters[2] = {
 133                         {
 134                                 .init = &lzma_lzma_decoder_init,
 135                                 .options = &coder->options,
 136                         }, {
 137                                 .init = NULL,
 138                         }
 139                 };
 140
 141                 const lzma_ret ret = lzma_next_filter_init(&coder->next,
 142                                 allocator, filters);
 143                 if (ret != LZMA_OK)
 144                         return ret;
 145
 146                 // Use a hack to set the uncompressed size.
 147                 lzma_lz_decoder_uncompressed(coder->next.coder,
 148                                 coder->uncompressed_size);
 149
 150                 coder->sequence = SEQ_CODE;
 151                 break;
 152         }
 153
 154         case SEQ_CODE: {
 155                 return coder->next.code(coder->next.coder,
 156                                 allocator, in, in_pos, in_size,
 157                                 out, out_pos, out_size, action);
 158         }
 159
 160         default:
 161                 return LZMA_PROG_ERROR;
 162         }
 163
 164         return LZMA_OK;
 165 }
 166
 167
 168 static void
 169 alone_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 170 {
 171         lzma_next_end(&coder->next, allocator);
 172         lzma_free(coder, allocator);
 173         return;
 174 }
 175
 176
 177 static lzma_ret
 178 alone_decoder_memconfig(lzma_coder *coder, uint64_t *memusage,
 179                 uint64_t *old_memlimit, uint64_t new_memlimit)
 180 {
 181         *memusage = coder->memusage;
 182         *old_memlimit = coder->memlimit;
 183
 184         if (new_memlimit != 0) {
 185                 if (new_memlimit < coder->memusage)
 186                         return LZMA_MEMLIMIT_ERROR;
 187
 188                 coder->memlimit = new_memlimit;
 189         }
 190
 191         return LZMA_OK;
 192 }
 193
 194
 195 extern lzma_ret
 196 lzma_alone_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 197                 uint64_t memlimit, bool picky)
 198 {
 199         lzma_next_coder_init(&lzma_alone_decoder_init, next, allocator);
 200
 201         if (memlimit == 0)
 202                 return LZMA_PROG_ERROR;
 203
 204         if (next->coder == NULL) {
 205                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 206                 if (next->coder == NULL)
 207                         return LZMA_MEM_ERROR;
 208
 209                 next->code = &alone_decode;
 210                 next->end = &alone_decoder_end;
 211                 next->memconfig = &alone_decoder_memconfig;
 212                 next->coder->next = LZMA_NEXT_CODER_INIT;
 213         }
 214
 215         next->coder->sequence = SEQ_PROPERTIES;
 216         next->coder->picky = picky;
 217         next->coder->pos = 0;
 218         next->coder->options.dict_size = 0;
 219         next->coder->options.preset_dict = NULL;
 220         next->coder->options.preset_dict_size = 0;
 221         next->coder->uncompressed_size = 0;
 222         next->coder->memlimit = memlimit;
 223         next->coder->memusage = LZMA_MEMUSAGE_BASE;
 224
 225         return LZMA_OK;
 226 }
 227
 228
 229 extern LZMA_API(lzma_ret)
 230 lzma_alone_decoder(lzma_stream *strm, uint64_t memlimit)
 231 {
 232         lzma_next_strm_init(lzma_alone_decoder_init, strm, memlimit, false);
 233
 234         strm->internal->supported_actions[LZMA_RUN] = true;
 235         strm->internal->supported_actions[LZMA_FINISH] = true;
 236
 237         return LZMA_OK;
 238 }