src/liblzma/simple/simple_coder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       simple_coder.c
   4 /// \brief      Wrapper for simple filters
   5 ///
   6 /// Simple filters don't change the size of the data i.e. number of bytes
   7 /// in equals the number of bytes out.
   8 //
   9 //  Copyright (C) 2007 Lasse Collin
  10 //
  11 //  This library is free software; you can redistribute it and/or
  12 //  modify it under the terms of the GNU Lesser General Public
  13 //  License as published by the Free Software Foundation; either
  14 //  version 2.1 of the License, or (at your option) any later version.
  15 //
  16 //  This library is distributed in the hope that it will be useful,
  17 //  but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 //  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 //  Lesser General Public License for more details.
  20 //
  21 ///////////////////////////////////////////////////////////////////////////////
  22
  23 #include "simple_private.h"
  24
  25
  26 /// Copied or encodes/decodes more data to out[]. Checks and updates
  27 /// uncompressed_size when we are the last coder in the chain.
  28 /// If we aren't the last filter in the chain, we don't need to care about
  29 /// uncompressed size, since we don't change it; the next filter in the
  30 /// chain will check it anyway.
  31 static lzma_ret
  32 copy_or_code(lzma_coder *coder, lzma_allocator *allocator,
  33                 const uint8_t *restrict in, size_t *restrict in_pos,
  34                 size_t in_size, uint8_t *restrict out,
  35                 size_t *restrict out_pos, size_t out_size, lzma_action action)
  36 {
  37         assert(!coder->end_was_reached);
  38
  39         if (coder->next.code == NULL) {
  40                 const size_t in_avail = in_size - *in_pos;
  41
  42                 if (!coder->is_encoder) {
  43                         // Limit in_size so that we don't copy too much.
  44                         if ((lzma_vli)(in_avail) > coder->uncompressed_size)
  45                                 in_size = *in_pos + (size_t)(
  46                                                 coder->uncompressed_size);
  47                 }
  48
  49                 const size_t out_start = *out_pos;
  50                 bufcpy(in, in_pos, in_size, out, out_pos, out_size);
  51
  52                 // Check if end of stream was reached.
  53                 if (coder->is_encoder) {
  54                         if (action == LZMA_FINISH && *in_pos == in_size)
  55                                 coder->end_was_reached = true;
  56                 } else if (coder->uncompressed_size
  57                                 != LZMA_VLI_VALUE_UNKNOWN) {
  58                         coder->uncompressed_size -= *out_pos - out_start;
  59                         if (coder->uncompressed_size == 0)
  60                                 coder->end_was_reached = true;
  61                 }
  62
  63         } else {
  64                 // Call the next coder in the chain to provide us some data.
  65                 // We don't care about uncompressed_size here, because
  66                 // the next filter in the chain will do it for us (since
  67                 // we don't change the size of the data).
  68                 const lzma_ret ret = coder->next.code(
  69                                 coder->next.coder, allocator,
  70                                 in, in_pos, in_size,
  71                                 out, out_pos, out_size, action);
  72
  73                 if (ret == LZMA_STREAM_END) {
  74                         assert(!coder->is_encoder
  75                                         || action == LZMA_FINISH);
  76                         coder->end_was_reached = true;
  77
  78                 } else if (ret != LZMA_OK) {
  79                         return ret;
  80                 }
  81         }
  82
  83         return LZMA_OK;
  84 }
  85
  86
  87 static size_t
  88 call_filter(lzma_coder *coder, uint8_t *buffer, size_t size)
  89 {
  90         const size_t filtered = coder->filter(coder->simple,
  91                         coder->now_pos, coder->is_encoder,
  92                         buffer, size);
  93         coder->now_pos += filtered;
  94         return filtered;
  95 }
  96
  97
  98 static lzma_ret
  99 simple_code(lzma_coder *coder, lzma_allocator *allocator,
 100                 const uint8_t *restrict in, size_t *restrict in_pos,
 101                 size_t in_size, uint8_t *restrict out,
 102                 size_t *restrict out_pos, size_t out_size, lzma_action action)
 103 {
 104         // Flush already filtered data from coder->buffer[] to out[].
 105         if (coder->pos < coder->filtered) {
 106                 bufcpy(coder->buffer, &coder->pos, coder->filtered,
 107                                 out, out_pos, out_size);
 108
 109                 // If we couldn't flush all the filtered data, return to
 110                 // application immediatelly.
 111                 if (coder->pos < coder->filtered)
 112                         return LZMA_OK;
 113
 114                 if (coder->end_was_reached) {
 115                         assert(coder->filtered == coder->size);
 116                         return LZMA_STREAM_END;
 117                 }
 118         }
 119
 120         // If we get here, there is no filtered data left in the buffer.
 121         coder->filtered = 0;
 122
 123         assert(!coder->end_was_reached);
 124
 125         // If there is more output space left than there is unfiltered data
 126         // in coder->buffer[], flush coder->buffer[] to out[], and copy/code
 127         // more data to out[] hopefully filling it completely. Then filter
 128         // the data in out[]. This step is where most of the data gets
 129         // filtered if the buffer sizes used by the application are reasonable.
 130         const size_t out_avail = out_size - *out_pos;
 131         const size_t buf_avail = coder->size - coder->pos;
 132         if (out_avail > buf_avail) {
 133                 // Store the old position so that we know from which byte
 134                 // to start filtering.
 135                 const size_t out_start = *out_pos;
 136
 137                 // Flush data from coder->buffer[] to out[], but don't reset
 138                 // coder->pos and coder->size yet. This way the coder can be
 139                 // restarted if the next filter in the chain returns e.g.
 140                 // LZMA_MEM_ERROR.
 141                 memcpy(out + *out_pos, coder->buffer + coder->pos, buf_avail);
 142                 *out_pos += buf_avail;
 143
 144                 // Copy/Encode/Decode more data to out[].
 145                 {
 146                         const lzma_ret ret = copy_or_code(coder, allocator,
 147                                         in, in_pos, in_size,
 148                                         out, out_pos, out_size, action);
 149                         assert(ret != LZMA_STREAM_END);
 150                         if (ret != LZMA_OK)
 151                                 return ret;
 152                 }
 153
 154                 // Filter out[].
 155                 const size_t size = *out_pos - out_start;
 156                 const size_t filtered = call_filter(
 157                                 coder, out + out_start, size);
 158
 159                 const size_t unfiltered = size - filtered;
 160                 assert(unfiltered <= coder->allocated / 2);
 161
 162                 // Now we can update coder->pos and coder->size, because
 163                 // the next coder in the chain (if any) was successful.
 164                 coder->pos = 0;
 165                 coder->size = unfiltered;
 166
 167                 if (coder->end_was_reached) {
 168                         // The last byte has been copied to out[] already.
 169                         // They are left as is.
 170                         coder->size = 0;
 171
 172                 } else if (unfiltered > 0) {
 173                         // There is unfiltered data left in out[]. Copy it to
 174                         // coder->buffer[] and rewind *out_pos appropriately.
 175                         *out_pos -= unfiltered;
 176                         memcpy(coder->buffer, out + *out_pos, unfiltered);
 177                 }
 178         } else if (coder->pos > 0) {
 179                 memmove(coder->buffer, coder->buffer + coder->pos, buf_avail);
 180                 coder->size -= coder->pos;
 181                 coder->pos = 0;
 182         }
 183
 184         assert(coder->pos == 0);
 185
 186         // If coder->buffer[] isn't empty, try to fill it by copying/decoding
 187         // more data. Then filter coder->buffer[] and copy the successfully
 188         // filtered data to out[]. It is probable, that some filtered and
 189         // unfiltered data will be left to coder->buffer[].
 190         if (coder->size > 0) {
 191                 {
 192                         const lzma_ret ret = copy_or_code(coder, allocator,
 193                                         in, in_pos, in_size,
 194                                         coder->buffer, &coder->size,
 195                                         coder->allocated, action);
 196                         assert(ret != LZMA_STREAM_END);
 197                         if (ret != LZMA_OK)
 198                                 return ret;
 199                 }
 200
 201                 coder->filtered = call_filter(
 202                                 coder, coder->buffer, coder->size);
 203
 204                 // Everything is considered to be filtered if coder->buffer[]
 205                 // contains the last bytes of the data.
 206                 if (coder->end_was_reached)
 207                         coder->filtered = coder->size;
 208
 209                 // Flush as much as possible.
 210                 bufcpy(coder->buffer, &coder->pos, coder->filtered,
 211                                 out, out_pos, out_size);
 212         }
 213
 214         // Check if we got everything done.
 215         if (coder->end_was_reached && coder->pos == coder->size)
 216                 return LZMA_STREAM_END;
 217
 218         return LZMA_OK;
 219 }
 220
 221
 222 static void
 223 simple_coder_end(lzma_coder *coder, lzma_allocator *allocator)
 224 {
 225         lzma_next_coder_end(&coder->next, allocator);
 226         lzma_free(coder->simple, allocator);
 227         lzma_free(coder, allocator);
 228         return;
 229 }
 230
 231
 232 extern lzma_ret
 233 lzma_simple_coder_init(lzma_next_coder *next, lzma_allocator *allocator,
 234                 const lzma_filter_info *filters,
 235                 size_t (*filter)(lzma_simple *simple, uint32_t now_pos,
 236                         bool is_encoder, uint8_t *buffer, size_t size),
 237                 size_t simple_size, size_t unfiltered_max, bool is_encoder)
 238 {
 239         // Allocate memory for the lzma_coder structure if needed.
 240         if (next->coder == NULL) {
 241                 // Here we allocate space also for the temporary buffer. We
 242                 // need twice the size of unfiltered_max, because then it
 243                 // is always possible to filter at least unfiltered_max bytes
 244                 // more data in coder->buffer[] if it can be filled completely.
 245                 next->coder = lzma_alloc(sizeof(lzma_coder)
 246                                 + 2 * unfiltered_max, allocator);
 247                 if (next->coder == NULL)
 248                         return LZMA_MEM_ERROR;
 249
 250                 next->code = &simple_code;
 251                 next->end = &simple_coder_end;
 252
 253                 next->coder->next = LZMA_NEXT_CODER_INIT;
 254                 next->coder->filter = filter;
 255                 next->coder->allocated = 2 * unfiltered_max;
 256
 257                 // Allocate memory for filter-specific data structure.
 258                 if (simple_size > 0) {
 259                         next->coder->simple = lzma_alloc(
 260                                         simple_size, allocator);
 261                         if (next->coder->simple == NULL)
 262                                 return LZMA_MEM_ERROR;
 263                 } else {
 264                         next->coder->simple = NULL;
 265                 }
 266         }
 267
 268         if (filters[0].options != NULL) {
 269                 const lzma_options_simple *simple = filters[0].options;
 270                 next->coder->now_pos = simple->start_offset;
 271         } else {
 272                 next->coder->now_pos = 0;
 273         }
 274
 275         // Reset variables.
 276         next->coder->is_encoder = is_encoder;
 277         next->coder->end_was_reached = false;
 278         next->coder->uncompressed_size = filters[0].uncompressed_size;
 279         next->coder->pos = 0;
 280         next->coder->filtered = 0;
 281         next->coder->size = 0;
 282
 283         return lzma_next_filter_init(
 284                         &next->coder->next, allocator, filters + 1);
 285 }