src/xzdec/xzdec.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       xzdec.c
   4 /// \brief      Simple single-threaded tool to uncompress .xz or .lzma files
   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 "sysdefs.h"
  21
  22 #include <errno.h>
  23 #include <stdio.h>
  24 #include <unistd.h>
  25
  26 #ifdef WIN32
  27 #       include <fcntl.h>
  28 #endif
  29
  30 #include "getopt.h"
  31 #include "physmem.h"
  32
  33
  34 #ifdef LZMADEC
  35 #       define TOOL_FORMAT "lzma"
  36 #else
  37 #       define TOOL_FORMAT "xz"
  38 #endif
  39
  40
  41 /// Number of bytes to use memory at maximum
  42 static uint64_t memlimit;
  43
  44 /// Program name to be shown in error messages
  45 static const char *argv0;
  46
  47
  48 static void lzma_attribute((noreturn))
  49 my_exit(void)
  50 {
  51         int status = EXIT_SUCCESS;
  52
  53         // Close stdout. We don't care about stderr, because we write to it
  54         // only when an error has already occurred.
  55         const int ferror_err = ferror(stdout);
  56         const int fclose_err = fclose(stdout);
  57
  58         if (ferror_err || fclose_err) {
  59                 // If it was fclose() that failed, we have the reason
  60                 // in errno. If only ferror() indicated an error,
  61                 // we have no idea what the reason was.
  62                 fprintf(stderr, "%s: Cannot write to standard output: %s\n",
  63                                 argv0, fclose_err
  64                                         ? strerror(errno) : "Unknown error");
  65                 status = EXIT_FAILURE;
  66         }
  67
  68         exit(status);
  69 }
  70
  71
  72 static void lzma_attribute((noreturn))
  73 help(void)
  74 {
  75         printf(
  76 "Usage: %s [OPTION]... [FILE]...\n"
  77 "Uncompress files in the ." TOOL_FORMAT " format to the standard output.\n"
  78 "\n"
  79 "  -c, --stdout       (ignored)\n"
  80 "  -d, --decompress   (ignored)\n"
  81 "  -k, --keep         (ignored)\n"
  82 "  -f, --force        (ignored)\n"
  83 "  -M, --memory=NUM   use NUM bytes of memory at maximum (0 means default);\n"
  84 "                     the suffixes k, M, G, Ki, Mi, and Gi are supported.\n"
  85 "  -h, --help         display this help and exit\n"
  86 "  -V, --version      display version and license information and exit\n"
  87 "\n"
  88 "With no FILE, or when FILE is -, read standard input.\n"
  89 "\n"
  90 "On this configuration, the tool will use about %" PRIu64
  91                 " MiB of memory at maximum.\n"
  92 "\n"
  93 "Report bugs to <" PACKAGE_BUGREPORT "> (in English or Finnish).\n",
  94                 argv0, (memlimit + 512 * 1024) / (1024 * 1024));
  95         my_exit();
  96 }
  97
  98
  99 static void lzma_attribute((noreturn))
 100 version(void)
 101 {
 102         printf(TOOL_FORMAT "dec " PACKAGE_VERSION "\n"
 103                         "liblzma %s\n", lzma_version_string());
 104
 105         my_exit();
 106 }
 107
 108
 109 /// Finds out the amount of physical memory in the system, and sets
 110 /// a default memory usage limit.
 111 static void
 112 set_default_memlimit(void)
 113 {
 114         const uint64_t mem = physmem();
 115
 116         if (mem == 0)
 117                 // Cannot autodetect, use 10 MiB as the default limit.
 118                 memlimit = (1U << 23) + (1U << 21);
 119         else
 120                 // Limit is 33 % of RAM.
 121                 memlimit = mem / 3;
 122
 123         return;
 124 }
 125
 126
 127 /// \brief      Converts a string to uint64_t
 128 ///
 129 /// This is rudely copied from src/xz/util.c and modified a little. :-(
 130 ///
 131 static uint64_t
 132 str_to_uint64(const char *value)
 133 {
 134         uint64_t result = 0;
 135
 136         if (*value < '0' || *value > '9') {
 137                 fprintf(stderr, "%s: %s: Not a number", argv0, value);
 138                 exit(EXIT_FAILURE);
 139         }
 140
 141         do {
 142                 // Don't overflow.
 143                 if (result > (UINT64_MAX - 9) / 10)
 144                         return UINT64_MAX;
 145
 146                 result *= 10;
 147                 result += *value - '0';
 148                 ++value;
 149         } while (*value >= '0' && *value <= '9');
 150
 151         if (*value != '\0') {
 152                 // Look for suffix.
 153                 static const struct {
 154                         const char name[4];
 155                         uint32_t multiplier;
 156                 } suffixes[] = {
 157                         { "k",   1000 },
 158                         { "kB",  1000 },
 159                         { "M",   1000000 },
 160                         { "MB",  1000000 },
 161                         { "G",   1000000000 },
 162                         { "GB",  1000000000 },
 163                         { "Ki",  1024 },
 164                         { "KiB", 1024 },
 165                         { "Mi",  1048576 },
 166                         { "MiB", 1048576 },
 167                         { "Gi",  1073741824 },
 168                         { "GiB", 1073741824 }
 169                 };
 170
 171                 uint32_t multiplier = 0;
 172                 for (size_t i = 0; i < ARRAY_SIZE(suffixes); ++i) {
 173                         if (strcmp(value, suffixes[i].name) == 0) {
 174                                 multiplier = suffixes[i].multiplier;
 175                                 break;
 176                         }
 177                 }
 178
 179                 if (multiplier == 0) {
 180                         fprintf(stderr, "%s: %s: Invalid suffix",
 181                                         argv0, value);
 182                         exit(EXIT_FAILURE);
 183                 }
 184
 185                 // Don't overflow here either.
 186                 if (result > UINT64_MAX / multiplier)
 187                         result = UINT64_MAX;
 188                 else
 189                         result *= multiplier;
 190         }
 191
 192         return result;
 193 }
 194
 195
 196 /// Parses command line options.
 197 static void
 198 parse_options(int argc, char **argv)
 199 {
 200         static const char short_opts[] = "cdkfM:hV";
 201         static const struct option long_opts[] = {
 202                 { "stdout",       no_argument,         NULL, 'c' },
 203                 { "to-stdout",    no_argument,         NULL, 'c' },
 204                 { "decompress",   no_argument,         NULL, 'd' },
 205                 { "uncompress",   no_argument,         NULL, 'd' },
 206                 { "force",        no_argument,         NULL, 'f' },
 207                 { "keep",         no_argument,         NULL, 'k' },
 208                 { "memory",       required_argument,   NULL, 'M' },
 209                 { "help",         no_argument,         NULL, 'h' },
 210                 { "version",      no_argument,         NULL, 'V' },
 211                 { NULL,           0,                   NULL, 0   }
 212         };
 213
 214         int c;
 215
 216         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 217                         != -1) {
 218                 switch (c) {
 219                 case 'c':
 220                 case 'd':
 221                 case 'f':
 222                 case 'k':
 223                         break;
 224
 225                 case 'M':
 226                         memlimit = str_to_uint64(optarg);
 227                         if (memlimit == 0)
 228                                 set_default_memlimit();
 229
 230                         break;
 231
 232                 case 'h':
 233                         help();
 234
 235                 case 'V':
 236                         version();
 237
 238                 default:
 239                         exit(EXIT_FAILURE);
 240                 }
 241         }
 242
 243         return;
 244 }
 245
 246
 247 static void
 248 uncompress(lzma_stream *strm, FILE *file, const char *filename)
 249 {
 250         lzma_ret ret;
 251
 252         // Initialize the decoder
 253 #ifdef LZMADEC
 254         ret = lzma_alone_decoder(strm, memlimit);
 255 #else
 256         ret = lzma_stream_decoder(strm, memlimit, LZMA_CONCATENATED);
 257 #endif
 258
 259         // The only reasonable error here is LZMA_MEM_ERROR.
 260         // FIXME: Maybe also LZMA_MEMLIMIT_ERROR in future?
 261         if (ret != LZMA_OK) {
 262                 fprintf(stderr, "%s: ", argv0);
 263
 264                 if (ret == LZMA_MEM_ERROR)
 265                         fprintf(stderr, "%s\n", strerror(ENOMEM));
 266                 else
 267                         fprintf(stderr, "Internal program error (bug)\n");
 268
 269                 exit(EXIT_FAILURE);
 270         }
 271
 272         // Input and output buffers
 273         uint8_t in_buf[BUFSIZ];
 274         uint8_t out_buf[BUFSIZ];
 275
 276         strm->avail_in = 0;
 277         strm->next_out = out_buf;
 278         strm->avail_out = BUFSIZ;
 279
 280         lzma_action action = LZMA_RUN;
 281
 282         while (true) {
 283                 if (strm->avail_in == 0) {
 284                         strm->next_in = in_buf;
 285                         strm->avail_in = fread(in_buf, 1, BUFSIZ, file);
 286
 287                         if (ferror(file)) {
 288                                 // POSIX says that fread() sets errno if
 289                                 // an error occurred. ferror() doesn't
 290                                 // touch errno.
 291                                 fprintf(stderr, "%s: %s: Error reading "
 292                                                 "input file: %s\n",
 293                                                 argv0, filename,
 294                                                 strerror(errno));
 295                                 exit(EXIT_FAILURE);
 296                         }
 297
 298 #ifndef LZMADEC
 299                         // When using LZMA_CONCATENATED, we need to tell
 300                         // liblzma when it has got all the input.
 301                         if (feof(file))
 302                                 action = LZMA_FINISH;
 303 #endif
 304                 }
 305
 306                 ret = lzma_code(strm, action);
 307
 308                 // Write and check write error before checking decoder error.
 309                 // This way as much data as possible gets written to output
 310                 // even if decoder detected an error.
 311                 if (strm->avail_out == 0 || ret != LZMA_OK) {
 312                         const size_t write_size = BUFSIZ - strm->avail_out;
 313
 314                         if (fwrite(out_buf, 1, write_size, stdout)
 315                                         != write_size) {
 316                                 // Wouldn't be a surprise if writing to stderr
 317                                 // would fail too but at least try to show an
 318                                 // error message.
 319                                 fprintf(stderr, "%s: Cannot write to "
 320                                                 "standard output: %s\n", argv0,
 321                                                 strerror(errno));
 322                                 exit(EXIT_FAILURE);
 323                         }
 324
 325                         strm->next_out = out_buf;
 326                         strm->avail_out = BUFSIZ;
 327                 }
 328
 329                 if (ret != LZMA_OK) {
 330                         if (ret == LZMA_STREAM_END) {
 331 #ifdef LZMADEC
 332                                 // Check that there's no trailing garbage.
 333                                 if (strm->avail_in != 0
 334                                                 || fread(in_buf, 1, 1, file)
 335                                                         != 0
 336                                                 || !feof(file))
 337                                         ret = LZMA_DATA_ERROR;
 338                                 else
 339                                         return;
 340 #else
 341                                 // lzma_stream_decoder() already guarantees
 342                                 // that there's no trailing garbage.
 343                                 assert(strm->avail_in == 0);
 344                                 assert(action == LZMA_FINISH);
 345                                 assert(feof(file));
 346                                 return;
 347 #endif
 348                         }
 349
 350                         const char *msg;
 351                         switch (ret) {
 352                         case LZMA_MEM_ERROR:
 353                                 msg = strerror(ENOMEM);
 354                                 break;
 355
 356                         case LZMA_MEMLIMIT_ERROR:
 357                                 msg = "Memory usage limit reached";
 358                                 break;
 359
 360                         case LZMA_FORMAT_ERROR:
 361                                 msg = "File format not recognized";
 362                                 break;
 363
 364                         case LZMA_OPTIONS_ERROR:
 365                                 // FIXME: Better message?
 366                                 msg = "Unsupported compression options";
 367                                 break;
 368
 369                         case LZMA_DATA_ERROR:
 370                                 msg = "File is corrupt";
 371                                 break;
 372
 373                         case LZMA_BUF_ERROR:
 374                                 msg = "Unexpected end of input";
 375                                 break;
 376
 377                         default:
 378                                 msg = "Internal program error (bug)";
 379                                 break;
 380                         }
 381
 382                         fprintf(stderr, "%s: %s: %s", argv0, filename, msg);
 383
 384                         exit(EXIT_FAILURE);
 385                 }
 386         }
 387 }
 388
 389
 390 int
 391 main(int argc, char **argv)
 392 {
 393         // Set the argv0 global so that we can print the command name in
 394         // error and help messages.
 395         argv0 = argv[0];
 396
 397         // Detect amount of installed RAM and set the memory usage limit.
 398         // This is needed before parsing the command line arguments.
 399         set_default_memlimit();
 400
 401         // Parse the command line options.
 402         parse_options(argc, argv);
 403
 404         // The same lzma_stream is used for all files that we decode. This way
 405         // we don't need to reallocate memory for every file if they use same
 406         // compression settings.
 407         lzma_stream strm = LZMA_STREAM_INIT;
 408
 409         // Some systems require setting stdin and stdout to binary mode.
 410 #ifdef WIN32
 411         setmode(fileno(stdin), O_BINARY);
 412         setmode(fileno(stdout), O_BINARY);
 413 #endif
 414
 415         if (optind == argc) {
 416                 // No filenames given, decode from stdin.
 417                 uncompress(&strm, stdin, "(stdin)");
 418         } else {
 419                 // Loop through the filenames given on the command line.
 420                 do {
 421                         // "-" indicates stdin.
 422                         if (strcmp(argv[optind], "-") == 0) {
 423                                 uncompress(&strm, stdin, "(stdin)");
 424                         } else {
 425                                 FILE *file = fopen(argv[optind], "rb");
 426                                 if (file == NULL) {
 427                                         fprintf(stderr, "%s: %s: %s\n",
 428                                                         argv0, argv[optind],
 429                                                         strerror(errno));
 430                                         exit(EXIT_FAILURE);
 431                                 }
 432
 433                                 uncompress(&strm, file, argv[optind]);
 434                                 fclose(file);
 435                         }
 436                 } while (++optind < argc);
 437         }
 438
 439 #ifndef NDEBUG
 440         // Free the memory only when debugging. Freeing wastes some time,
 441         // but allows detecting possible memory leaks with Valgrind.
 442         lzma_end(&strm);
 443 #endif
 444
 445         my_exit();
 446 }