src/liblzma/common/index_encoder.c

   1 // SPDX-License-Identifier: 0BSD
   2
   3 ///////////////////////////////////////////////////////////////////////////////
   4 //
   5 /// \file       index_encoder.c
   6 /// \brief      Encodes the Index field
   7 //
   8 //  Author:     Lasse Collin
   9 //
  10 ///////////////////////////////////////////////////////////////////////////////
  11
  12 #include "index_encoder.h"
  13 #include "index.h"
  14 #include "check.h"
  15
  16
  17 typedef struct {
  18         enum {
  19                 SEQ_INDICATOR,
  20                 SEQ_COUNT,
  21                 SEQ_UNPADDED,
  22                 SEQ_UNCOMPRESSED,
  23                 SEQ_NEXT,
  24                 SEQ_PADDING,
  25                 SEQ_CRC32,
  26         } sequence;
  27
  28         /// Index being encoded
  29         const lzma_index *index;
  30
  31         /// Iterator for the Index being encoded
  32         lzma_index_iter iter;
  33
  34         /// Position in integers
  35         size_t pos;
  36
  37         /// CRC32 of the List of Records field
  38         uint32_t crc32;
  39 } lzma_index_coder;
  40
  41
  42 static lzma_ret
  43 index_encode(void *coder_ptr,
  44                 const lzma_allocator *allocator lzma_attribute((__unused__)),
  45                 const uint8_t *restrict in lzma_attribute((__unused__)),
  46                 size_t *restrict in_pos lzma_attribute((__unused__)),
  47                 size_t in_size lzma_attribute((__unused__)),
  48                 uint8_t *restrict out, size_t *restrict out_pos,
  49                 size_t out_size,
  50                 lzma_action action lzma_attribute((__unused__)))
  51 {
  52         lzma_index_coder *coder = coder_ptr;
  53
  54         // Position where to start calculating CRC32. The idea is that we
  55         // need to call lzma_crc32() only once per call to index_encode().
  56         const size_t out_start = *out_pos;
  57
  58         // Return value to use if we return at the end of this function.
  59         // We use "goto out" to jump out of the while-switch construct
  60         // instead of returning directly, because that way we don't need
  61         // to copypaste the lzma_crc32() call to many places.
  62         lzma_ret ret = LZMA_OK;
  63
  64         while (*out_pos < out_size)
  65         switch (coder->sequence) {
  66         case SEQ_INDICATOR:
  67                 out[*out_pos] = INDEX_INDICATOR;
  68                 ++*out_pos;
  69                 coder->sequence = SEQ_COUNT;
  70                 break;
  71
  72         case SEQ_COUNT: {
  73                 const lzma_vli count = lzma_index_block_count(coder->index);
  74                 ret = lzma_vli_encode(count, &coder->pos,
  75                                 out, out_pos, out_size);
  76                 if (ret != LZMA_STREAM_END)
  77                         goto out;
  78
  79                 ret = LZMA_OK;
  80                 coder->pos = 0;
  81                 coder->sequence = SEQ_NEXT;
  82                 break;
  83         }
  84
  85         case SEQ_NEXT:
  86                 if (lzma_index_iter_next(
  87                                 &coder->iter, LZMA_INDEX_ITER_BLOCK)) {
  88                         // Get the size of the Index Padding field.
  89                         coder->pos = lzma_index_padding_size(coder->index);
  90                         assert(coder->pos <= 3);
  91                         coder->sequence = SEQ_PADDING;
  92                         break;
  93                 }
  94
  95                 coder->sequence = SEQ_UNPADDED;
  96
  97         // Fall through
  98
  99         case SEQ_UNPADDED:
 100         case SEQ_UNCOMPRESSED: {
 101                 const lzma_vli size = coder->sequence == SEQ_UNPADDED
 102                                 ? coder->iter.block.unpadded_size
 103                                 : coder->iter.block.uncompressed_size;
 104
 105                 ret = lzma_vli_encode(size, &coder->pos,
 106                                 out, out_pos, out_size);
 107                 if (ret != LZMA_STREAM_END)
 108                         goto out;
 109
 110                 ret = LZMA_OK;
 111                 coder->pos = 0;
 112
 113                 // Advance to SEQ_UNCOMPRESSED or SEQ_NEXT.
 114                 ++coder->sequence;
 115                 break;
 116         }
 117
 118         case SEQ_PADDING:
 119                 if (coder->pos > 0) {
 120                         --coder->pos;
 121                         out[(*out_pos)++] = 0x00;
 122                         break;
 123                 }
 124
 125                 // Finish the CRC32 calculation.
 126                 coder->crc32 = lzma_crc32(out + out_start,
 127                                 *out_pos - out_start, coder->crc32);
 128
 129                 coder->sequence = SEQ_CRC32;
 130
 131         // Fall through
 132
 133         case SEQ_CRC32:
 134                 // We don't use the main loop, because we don't want
 135                 // coder->crc32 to be touched anymore.
 136                 do {
 137                         if (*out_pos == out_size)
 138                                 return LZMA_OK;
 139
 140                         out[*out_pos] = (coder->crc32 >> (coder->pos * 8))
 141                                         & 0xFF;
 142                         ++*out_pos;
 143
 144                 } while (++coder->pos < 4);
 145
 146                 return LZMA_STREAM_END;
 147
 148         default:
 149                 assert(0);
 150                 return LZMA_PROG_ERROR;
 151         }
 152
 153 out:
 154         // Update the CRC32.
 155         //
 156         // Avoid null pointer + 0 (undefined behavior) in "out + out_start".
 157         // In such a case we had no input and thus out_used == 0.
 158         {
 159                 const size_t out_used = *out_pos - out_start;
 160                 if (out_used > 0)
 161                         coder->crc32 = lzma_crc32(out + out_start,
 162                                         out_used, coder->crc32);
 163         }
 164
 165         return ret;
 166 }
 167
 168
 169 static void
 170 index_encoder_end(void *coder, const lzma_allocator *allocator)
 171 {
 172         lzma_free(coder, allocator);
 173         return;
 174 }
 175
 176
 177 static void
 178 index_encoder_reset(lzma_index_coder *coder, const lzma_index *i)
 179 {
 180         lzma_index_iter_init(&coder->iter, i);
 181
 182         coder->sequence = SEQ_INDICATOR;
 183         coder->index = i;
 184         coder->pos = 0;
 185         coder->crc32 = 0;
 186
 187         return;
 188 }
 189
 190
 191 extern lzma_ret
 192 lzma_index_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
 193                 const lzma_index *i)
 194 {
 195         lzma_next_coder_init(&lzma_index_encoder_init, next, allocator);
 196
 197         if (i == NULL)
 198                 return LZMA_PROG_ERROR;
 199
 200         if (next->coder == NULL) {
 201                 next->coder = lzma_alloc(sizeof(lzma_index_coder), allocator);
 202                 if (next->coder == NULL)
 203                         return LZMA_MEM_ERROR;
 204
 205                 next->code = &index_encode;
 206                 next->end = &index_encoder_end;
 207         }
 208
 209         index_encoder_reset(next->coder, i);
 210
 211         return LZMA_OK;
 212 }
 213
 214
 215 extern LZMA_API(lzma_ret)
 216 lzma_index_encoder(lzma_stream *strm, const lzma_index *i)
 217 {
 218         lzma_next_strm_init(lzma_index_encoder_init, strm, i);
 219
 220         strm->internal->supported_actions[LZMA_RUN] = true;
 221         strm->internal->supported_actions[LZMA_FINISH] = true;
 222
 223         return LZMA_OK;
 224 }
 225
 226
 227 extern LZMA_API(lzma_ret)
 228 lzma_index_buffer_encode(const lzma_index *i,
 229                 uint8_t *out, size_t *out_pos, size_t out_size)
 230 {
 231         // Validate the arguments.
 232         if (i == NULL || out == NULL || out_pos == NULL || *out_pos > out_size)
 233                 return LZMA_PROG_ERROR;
 234
 235         // Don't try to encode if there's not enough output space.
 236         if (out_size - *out_pos < lzma_index_size(i))
 237                 return LZMA_BUF_ERROR;
 238
 239         // The Index encoder needs just one small data structure so we can
 240         // allocate it on stack.
 241         lzma_index_coder coder;
 242         index_encoder_reset(&coder, i);
 243
 244         // Do the actual encoding. This should never fail, but store
 245         // the original *out_pos just in case.
 246         const size_t out_start = *out_pos;
 247         lzma_ret ret = index_encode(&coder, NULL, NULL, NULL, 0,
 248                         out, out_pos, out_size, LZMA_RUN);
 249
 250         if (ret == LZMA_STREAM_END) {
 251                 ret = LZMA_OK;
 252         } else {
 253                 // We should never get here, but just in case, restore the
 254                 // output position and set the error accordingly if something
 255                 // goes wrong and debugging isn't enabled.
 256                 assert(0);
 257                 *out_pos = out_start;
 258                 ret = LZMA_PROG_ERROR;
 259         }
 260
 261         return ret;
 262 }