src/xz/process.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       process.c
   4 /// \brief      Compresses or uncompresses a file
   5 //
   6 //  Copyright (C) 2007 Lasse Collin
   7 //
   8 //  This program 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 program 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 "private.h"
  21
  22
  23 enum operation_mode opt_mode = MODE_COMPRESS;
  24
  25 enum format_type opt_format = FORMAT_AUTO;
  26
  27
  28 /// Stream used to communicate with liblzma
  29 static lzma_stream strm = LZMA_STREAM_INIT;
  30
  31 /// Filters needed for all encoding all formats, and also decoding in raw data
  32 static lzma_filter filters[LZMA_FILTERS_MAX + 1];
  33
  34 /// Number of filters. Zero indicates that we are using a preset.
  35 static size_t filters_count = 0;
  36
  37 /// Number of the preset (1-9)
  38 static size_t preset_number = 7;
  39
  40 /// Indicate if no preset has been given. In that case, we will auto-adjust
  41 /// the compression preset so that it doesn't use too much RAM.
  42 // FIXME
  43 static bool preset_default = true;
  44
  45 /// Integrity check type
  46 static lzma_check check = LZMA_CHECK_CRC64;
  47
  48
  49 extern void
  50 coder_set_check(lzma_check new_check)
  51 {
  52         check = new_check;
  53         return;
  54 }
  55
  56
  57 extern void
  58 coder_set_preset(size_t new_preset)
  59 {
  60         preset_number = new_preset;
  61         preset_default = false;
  62         return;
  63 }
  64
  65
  66 extern void
  67 coder_add_filter(lzma_vli id, void *options)
  68 {
  69         if (filters_count == LZMA_FILTERS_MAX)
  70                 message_fatal(_("Maximum number of filters is four"));
  71
  72         filters[filters_count].id = id;
  73         filters[filters_count].options = options;
  74         ++filters_count;
  75
  76         return;
  77 }
  78
  79
  80 extern void
  81 coder_set_compression_settings(void)
  82 {
  83         // Options for LZMA1 or LZMA2 in case we are using a preset.
  84         static lzma_options_lzma opt_lzma;
  85
  86         if (filters_count == 0) {
  87                 // We are using a preset. This is not a good idea in raw mode
  88                 // except when playing around with things. Different versions
  89                 // of this software may use different options in presets, and
  90                 // thus make uncompressing the raw data difficult.
  91                 if (opt_format == FORMAT_RAW) {
  92                         // The message is shown only if warnings are allowed
  93                         // but the exit status isn't changed.
  94                         message(V_WARNING, _("Using a preset in raw mode "
  95                                         "is discouraged."));
  96                         message(V_WARNING, _("The exact options of the "
  97                                         "presets may vary between software "
  98                                         "versions."));
  99                 }
 100
 101                 // Get the preset for LZMA1 or LZMA2.
 102                 if (lzma_lzma_preset(&opt_lzma, preset_number))
 103                         message_bug();
 104
 105                 // Use LZMA2 except with --format=lzma we use LZMA1.
 106                 filters[0].id = opt_format == FORMAT_LZMA
 107                                 ? LZMA_FILTER_LZMA1 : LZMA_FILTER_LZMA2;
 108                 filters[0].options = &opt_lzma;
 109                 filters_count = 1;
 110         }
 111
 112         // Terminate the filter options array.
 113         filters[filters_count].id = LZMA_VLI_UNKNOWN;
 114
 115         // If we are using the LZMA_Alone format, allow exactly one filter
 116         // which has to be LZMA.
 117         if (opt_format == FORMAT_LZMA && (filters_count != 1
 118                         || filters[0].id != LZMA_FILTER_LZMA1))
 119                 message_fatal(_("With --format=lzma only the LZMA1 filter "
 120                                 "is supported"));
 121
 122         // TODO: liblzma probably needs an API to validate the filter chain.
 123
 124         // If using --format=raw, we can be decoding.
 125         uint64_t memory_usage;
 126         uint64_t memory_limit;
 127         if (opt_mode == MODE_COMPRESS) {
 128                 memory_usage = lzma_memusage_encoder(filters);
 129                 memory_limit = hardware_memlimit_encoder();
 130         } else {
 131                 memory_usage = lzma_memusage_decoder(filters);
 132                 memory_limit = hardware_memlimit_decoder();
 133         }
 134
 135         if (memory_usage == UINT64_MAX)
 136                 message_bug();
 137
 138         if (preset_default) {
 139                 // When no preset was explicitly requested, we use the default
 140                 // preset only if the memory usage limit allows. Otherwise we
 141                 // select a lower preset automatically.
 142                 while (memory_usage > memory_limit) {
 143                         if (preset_number == 1)
 144                                 message_fatal(_("Memory usage limit is too "
 145                                                 "small for any internal "
 146                                                 "filter preset"));
 147
 148                         if (lzma_lzma_preset(&opt_lzma, --preset_number))
 149                                 message_bug();
 150
 151                         memory_usage = lzma_memusage_encoder(filters);
 152                 }
 153         } else {
 154                 if (memory_usage > memory_limit)
 155                         message_fatal(_("Memory usage limit is too small "
 156                                         "for the given filter setup"));
 157         }
 158
 159         // Limit the number of worker threads so that memory usage
 160         // limit isn't exceeded.
 161         assert(memory_usage > 0);
 162         size_t thread_limit = memory_limit / memory_usage;
 163         if (thread_limit == 0)
 164                 thread_limit = 1;
 165
 166         if (opt_threads > thread_limit)
 167                 opt_threads = thread_limit;
 168
 169         return;
 170 }
 171
 172
 173 static bool
 174 coder_init(void)
 175 {
 176         lzma_ret ret = LZMA_PROG_ERROR;
 177
 178         if (opt_mode == MODE_COMPRESS) {
 179                 switch (opt_format) {
 180                 case FORMAT_AUTO:
 181                         // args.c ensures this.
 182                         assert(0);
 183                         break;
 184
 185                 case FORMAT_XZ:
 186                         ret = lzma_stream_encoder(&strm, filters, check);
 187                         break;
 188
 189                 case FORMAT_LZMA:
 190                         ret = lzma_alone_encoder(&strm, filters[0].options);
 191                         break;
 192
 193                 case FORMAT_RAW:
 194                         ret = lzma_raw_encoder(&strm, filters);
 195                         break;
 196                 }
 197         } else {
 198                 const uint32_t flags = LZMA_TELL_UNSUPPORTED_CHECK
 199                                 | LZMA_CONCATENATED;
 200
 201                 switch (opt_format) {
 202                 case FORMAT_AUTO:
 203                         ret = lzma_auto_decoder(&strm,
 204                                         hardware_memlimit_decoder(), flags);
 205                         break;
 206
 207                 case FORMAT_XZ:
 208                         ret = lzma_stream_decoder(&strm,
 209                                         hardware_memlimit_decoder(), flags);
 210                         break;
 211
 212                 case FORMAT_LZMA:
 213                         ret = lzma_alone_decoder(&strm,
 214                                         hardware_memlimit_decoder());
 215                         break;
 216
 217                 case FORMAT_RAW:
 218                         // Memory usage has already been checked in args.c.
 219                         // FIXME Comment
 220                         ret = lzma_raw_decoder(&strm, filters);
 221                         break;
 222                 }
 223         }
 224
 225         if (ret != LZMA_OK) {
 226                 if (ret == LZMA_MEM_ERROR)
 227                         message_error("%s", message_strm(LZMA_MEM_ERROR));
 228                 else
 229                         message_bug();
 230
 231                 return true;
 232         }
 233
 234         return false;
 235 }
 236
 237
 238 static bool
 239 coder_run(file_pair *pair)
 240 {
 241         // Buffers to hold input and output data.
 242         uint8_t in_buf[IO_BUFFER_SIZE];
 243         uint8_t out_buf[IO_BUFFER_SIZE];
 244
 245         // Initialize the progress indicator.
 246         const uint64_t in_size = pair->src_st.st_size <= (off_t)(0)
 247                         ? 0 : (uint64_t)(pair->src_st.st_size);
 248         message_progress_start(pair->src_name, in_size);
 249
 250         lzma_action action = LZMA_RUN;
 251         lzma_ret ret;
 252
 253         strm.avail_in = 0;
 254         strm.next_out = out_buf;
 255         strm.avail_out = IO_BUFFER_SIZE;
 256
 257         while (!user_abort) {
 258                 // Fill the input buffer if it is empty and we haven't reached
 259                 // end of file yet.
 260                 if (strm.avail_in == 0 && !pair->src_eof) {
 261                         strm.next_in = in_buf;
 262                         strm.avail_in = io_read(pair, in_buf, IO_BUFFER_SIZE);
 263
 264                         if (strm.avail_in == SIZE_MAX)
 265                                 break;
 266
 267                         // Encoder needs to know when we have given all the
 268                         // input to it. The decoders need to know it too when
 269                         // we are using LZMA_CONCATENATED.
 270                         if (pair->src_eof)
 271                                 action = LZMA_FINISH;
 272                 }
 273
 274                 // Let liblzma do the actual work.
 275                 ret = lzma_code(&strm, action);
 276
 277                 // Write out if the output buffer became full.
 278                 if (strm.avail_out == 0) {
 279                         if (opt_mode != MODE_TEST && io_write(pair, out_buf,
 280                                         IO_BUFFER_SIZE - strm.avail_out))
 281                                 return false;
 282
 283                         strm.next_out = out_buf;
 284                         strm.avail_out = IO_BUFFER_SIZE;
 285                 }
 286
 287                 if (ret != LZMA_OK) {
 288                         // Determine if the return value indicates that we
 289                         // won't continue coding.
 290                         const bool stop = ret != LZMA_NO_CHECK
 291                                         && ret != LZMA_UNSUPPORTED_CHECK;
 292
 293                         if (stop) {
 294                                 // First print the final progress info.
 295                                 // This way the user sees more accurately
 296                                 // where the error occurred. Note that we
 297                                 // print this *before* the possible error
 298                                 // message.
 299                                 //
 300                                 // FIXME: What if something goes wrong
 301                                 // after this?
 302                                 message_progress_end(strm.total_in,
 303                                                 strm.total_out,
 304                                                 ret == LZMA_STREAM_END);
 305
 306                                 // Write the remaining bytes even if something
 307                                 // went wrong, because that way the user gets
 308                                 // as much data as possible, which can be good
 309                                 // when trying to get at least some useful
 310                                 // data out of damaged files.
 311                                 if (opt_mode != MODE_TEST && io_write(pair,
 312                                                 out_buf, IO_BUFFER_SIZE
 313                                                         - strm.avail_out))
 314                                         return false;
 315                         }
 316
 317                         if (ret == LZMA_STREAM_END) {
 318                                 // Check that there is no trailing garbage.
 319                                 // This is needed for LZMA_Alone and raw
 320                                 // streams.
 321                                 if (strm.avail_in == 0 && (pair->src_eof
 322                                                 || io_read(pair, in_buf, 1)
 323                                                         == 0)) {
 324                                         assert(pair->src_eof);
 325                                         return true;
 326                                 }
 327
 328                                 // FIXME: What about io_read() failing?
 329
 330                                 // We hadn't reached the end of the file.
 331                                 ret = LZMA_DATA_ERROR;
 332                                 assert(stop);
 333                         }
 334
 335                         // If we get here and stop is true, something went
 336                         // wrong and we print an error. Otherwise it's just
 337                         // a warning and coding can continue.
 338                         if (stop) {
 339                                 message_error("%s: %s", pair->src_name,
 340                                                 message_strm(ret));
 341                         } else {
 342                                 message_warning("%s: %s", pair->src_name,
 343                                                 message_strm(ret));
 344
 345                                 // When compressing, all possible errors set
 346                                 // stop to true.
 347                                 assert(opt_mode != MODE_COMPRESS);
 348                         }
 349
 350                         if (ret == LZMA_MEMLIMIT_ERROR) {
 351                                 // Figure out how much memory would have
 352                                 // actually needed.
 353                                 // TODO
 354                         }
 355
 356                         if (stop)
 357                                 return false;
 358                 }
 359
 360                 // Show progress information if --verbose was specified and
 361                 // stderr is a terminal.
 362                 message_progress_update(strm.total_in, strm.total_out);
 363         }
 364
 365         return false;
 366 }
 367
 368
 369 extern void
 370 process_file(const char *filename)
 371 {
 372         // First try initializing the coder. If it fails, it's useless to try
 373         // opening the file. Check also for user_abort just in case if we had
 374         // got a signal while initializing the coder.
 375         if (coder_init() || user_abort)
 376                 return;
 377
 378         // Try to open the input and output files.
 379         file_pair *pair = io_open(filename);
 380         if (pair == NULL)
 381                 return;
 382
 383         // Do the actual coding.
 384         const bool success = coder_run(pair);
 385
 386         // Close the file pair. It needs to know if coding was successful to
 387         // know if the source or target file should be unlinked.
 388         io_close(pair, success);
 389
 390         return;
 391 }