src/xzdec/xzdec.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       xzdec.c
   4 /// \brief      Simple single-threaded tool to uncompress .xz or .lzma 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 "sysdefs.h"
  14 #include "lzma.h"
  15
  16 #include <stdarg.h>
  17 #include <errno.h>
  18 #include <stdio.h>
  19 #include <unistd.h>
  20
  21 #ifdef DOSLIKE
  22 #       include <fcntl.h>
  23 #       include <io.h>
  24 #endif
  25
  26 #include "getopt.h"
  27 #include "physmem.h"
  28
  29
  30 #ifdef LZMADEC
  31 #       define TOOL_FORMAT "lzma"
  32 #else
  33 #       define TOOL_FORMAT "xz"
  34 #endif
  35
  36
  37 /// Number of bytes to use memory at maximum
  38 static uint64_t memlimit;
  39
  40 /// Error messages are suppressed if this is zero, which is the case when
  41 /// --quiet has been given at least twice.
  42 static unsigned int display_errors = 2;
  43
  44 /// Program name to be shown in error messages
  45 static const char *argv0;
  46
  47
  48 static void lzma_attribute((format(printf, 1, 2)))
  49 my_errorf(const char *fmt, ...)
  50 {
  51         va_list ap;
  52         va_start(ap, fmt);
  53
  54         if (display_errors) {
  55                 fprintf(stderr, "%s: ", argv0);
  56                 vfprintf(stderr, fmt, ap);
  57                 fprintf(stderr, "\n");
  58         }
  59
  60         va_end(ap);
  61         return;
  62 }
  63
  64
  65 static void lzma_attribute((noreturn))
  66 my_exit(void)
  67 {
  68         int status = EXIT_SUCCESS;
  69
  70         // Close stdout. We don't care about stderr, because we write to it
  71         // only when an error has already occurred.
  72         const int ferror_err = ferror(stdout);
  73         const int fclose_err = fclose(stdout);
  74
  75         if (ferror_err || fclose_err) {
  76                 // If it was fclose() that failed, we have the reason
  77                 // in errno. If only ferror() indicated an error,
  78                 // we have no idea what the reason was.
  79                 my_errorf("Writing to standard output failed: %s", fclose_err
  80                                 ? strerror(errno) : "Unknown error");
  81                 status = EXIT_FAILURE;
  82         }
  83
  84         exit(status);
  85 }
  86
  87
  88 static void lzma_attribute((noreturn))
  89 help(void)
  90 {
  91         printf(
  92 "Usage: %s [OPTION]... [FILE]...\n"
  93 "Uncompress files in the ." TOOL_FORMAT " format to the standard output.\n"
  94 "\n"
  95 "  -c, --stdout       (ignored)\n"
  96 "  -d, --decompress   (ignored)\n"
  97 "  -k, --keep         (ignored)\n"
  98 "  -M, --memory=NUM   use NUM bytes of memory at maximum (0 means default)\n"
  99 "  -q, --quiet        specify *twice* to suppress errors\n"
 100 "  -Q, --no-warn      (ignored)\n"
 101 "  -h, --help         display this help and exit\n"
 102 "  -V, --version      display the version number and exit\n"
 103 "\n"
 104 "With no FILE, or when FILE is -, read standard input.\n"
 105 "\n"
 106 "On this system and configuration, this program will use a maximum of roughly\n"
 107 "%" PRIu64 " MiB RAM.\n"
 108 "\n"
 109 "Report bugs to <" PACKAGE_BUGREPORT "> (in English or Finnish).\n"
 110 PACKAGE_NAME " home page: <" PACKAGE_HOMEPAGE ">\n",
 111                 argv0, memlimit / (1024 * 1024));
 112         my_exit();
 113 }
 114
 115
 116 static void lzma_attribute((noreturn))
 117 version(void)
 118 {
 119         printf(TOOL_FORMAT "dec (" PACKAGE_NAME ") " LZMA_VERSION_STRING "\n"
 120                         "liblzma %s\n", lzma_version_string());
 121
 122         my_exit();
 123 }
 124
 125
 126 /// Find out the amount of physical memory (RAM) in the system, and set
 127 /// the memory usage limit to the given percentage of RAM.
 128 static void
 129 memlimit_set_percentage(uint32_t percentage)
 130 {
 131         uint64_t mem = physmem();
 132
 133         // If we cannot determine the amount of RAM, assume 32 MiB.
 134         if (mem == 0)
 135                 mem = UINT64_C(32) * 1024 * 1024;
 136
 137         memlimit = percentage * mem / 100;
 138         return;
 139 }
 140
 141
 142 /// Set the memory usage limit to give number of bytes. Zero is a special
 143 /// value to indicate the default limit.
 144 static void
 145 memlimit_set(uint64_t new_memlimit)
 146 {
 147         if (new_memlimit == 0)
 148                 memlimit_set_percentage(40);
 149         else
 150                 memlimit = new_memlimit;
 151
 152         return;
 153 }
 154
 155
 156 /// \brief      Convert a string to uint64_t
 157 ///
 158 /// This is rudely copied from src/xz/util.c and modified a little. :-(
 159 ///
 160 /// \param      max     Return value when the string "max" was specified.
 161 ///
 162 static uint64_t
 163 str_to_uint64(const char *value, uint64_t max)
 164 {
 165         uint64_t result = 0;
 166
 167         // Accept special value "max".
 168         if (strcmp(value, "max") == 0)
 169                 return max;
 170
 171         if (*value < '0' || *value > '9') {
 172                 my_errorf("%s: Value is not a non-negative decimal integer",
 173                                 value);
 174                 exit(EXIT_FAILURE);
 175         }
 176
 177         do {
 178                 // Don't overflow.
 179                 if (result > (UINT64_MAX - 9) / 10)
 180                         return UINT64_MAX;
 181
 182                 result *= 10;
 183                 result += *value - '0';
 184                 ++value;
 185         } while (*value >= '0' && *value <= '9');
 186
 187         if (*value != '\0') {
 188                 // Look for suffix.
 189                 static const struct {
 190                         const char name[4];
 191                         uint32_t multiplier;
 192                 } suffixes[] = {
 193                         { "k",   1000 },
 194                         { "kB",  1000 },
 195                         { "M",   1000000 },
 196                         { "MB",  1000000 },
 197                         { "G",   1000000000 },
 198                         { "GB",  1000000000 },
 199                         { "Ki",  1024 },
 200                         { "KiB", 1024 },
 201                         { "Mi",  1048576 },
 202                         { "MiB", 1048576 },
 203                         { "Gi",  1073741824 },
 204                         { "GiB", 1073741824 }
 205                 };
 206
 207                 uint32_t multiplier = 0;
 208                 for (size_t i = 0; i < ARRAY_SIZE(suffixes); ++i) {
 209                         if (strcmp(value, suffixes[i].name) == 0) {
 210                                 multiplier = suffixes[i].multiplier;
 211                                 break;
 212                         }
 213                 }
 214
 215                 if (multiplier == 0) {
 216                         my_errorf("%s: Invalid suffix", value);
 217                         exit(EXIT_FAILURE);
 218                 }
 219
 220                 // Don't overflow here either.
 221                 if (result > UINT64_MAX / multiplier)
 222                         result = UINT64_MAX;
 223                 else
 224                         result *= multiplier;
 225         }
 226
 227         return result;
 228 }
 229
 230
 231 /// Parses command line options.
 232 static void
 233 parse_options(int argc, char **argv)
 234 {
 235         static const char short_opts[] = "cdkM:hqQV";
 236         static const struct option long_opts[] = {
 237                 { "stdout",       no_argument,         NULL, 'c' },
 238                 { "to-stdout",    no_argument,         NULL, 'c' },
 239                 { "decompress",   no_argument,         NULL, 'd' },
 240                 { "uncompress",   no_argument,         NULL, 'd' },
 241                 { "keep",         no_argument,         NULL, 'k' },
 242                 { "memory",       required_argument,   NULL, 'M' },
 243                 { "quiet",        no_argument,         NULL, 'q' },
 244                 { "no-warn",      no_argument,         NULL, 'Q' },
 245                 { "help",         no_argument,         NULL, 'h' },
 246                 { "version",      no_argument,         NULL, 'V' },
 247                 { NULL,           0,                   NULL, 0   }
 248         };
 249
 250         int c;
 251
 252         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 253                         != -1) {
 254                 switch (c) {
 255                 case 'c':
 256                 case 'd':
 257                 case 'k':
 258                 case 'Q':
 259                         break;
 260
 261                 case 'M': {
 262                         // Support specifying the limit as a percentage of
 263                         // installed physical RAM.
 264                         const size_t len = strlen(optarg);
 265                         if (len > 0 && optarg[len - 1] == '%') {
 266                                 // Memory limit is a percentage of total
 267                                 // installed RAM.
 268                                 optarg[len - 1] = '\0';
 269                                 const uint64_t percentage
 270                                                 = str_to_uint64(optarg, 100);
 271                                 if (percentage < 1 || percentage > 100) {
 272                                         my_errorf("Percentage must be in "
 273                                                         "the range [1, 100]");
 274                                         exit(EXIT_FAILURE);
 275                                 }
 276
 277                                 memlimit_set_percentage(percentage);
 278                         } else {
 279                                 memlimit_set(str_to_uint64(
 280                                                 optarg, UINT64_MAX));
 281                         }
 282
 283                         break;
 284                 }
 285
 286                 case 'q':
 287                         if (display_errors > 0)
 288                                 --display_errors;
 289
 290                         break;
 291
 292                 case 'h':
 293                         help();
 294
 295                 case 'V':
 296                         version();
 297
 298                 default:
 299                         exit(EXIT_FAILURE);
 300                 }
 301         }
 302
 303         return;
 304 }
 305
 306
 307 static void
 308 uncompress(lzma_stream *strm, FILE *file, const char *filename)
 309 {
 310         lzma_ret ret;
 311
 312         // Initialize the decoder
 313 #ifdef LZMADEC
 314         ret = lzma_alone_decoder(strm, memlimit);
 315 #else
 316         ret = lzma_stream_decoder(strm, memlimit, LZMA_CONCATENATED);
 317 #endif
 318
 319         // The only reasonable error here is LZMA_MEM_ERROR.
 320         // FIXME: Maybe also LZMA_MEMLIMIT_ERROR in future?
 321         if (ret != LZMA_OK) {
 322                 my_errorf("%s", ret == LZMA_MEM_ERROR ? strerror(ENOMEM)
 323                                 : "Internal error (bug)");
 324                 exit(EXIT_FAILURE);
 325         }
 326
 327         // Input and output buffers
 328         uint8_t in_buf[BUFSIZ];
 329         uint8_t out_buf[BUFSIZ];
 330
 331         strm->avail_in = 0;
 332         strm->next_out = out_buf;
 333         strm->avail_out = BUFSIZ;
 334
 335         lzma_action action = LZMA_RUN;
 336
 337         while (true) {
 338                 if (strm->avail_in == 0) {
 339                         strm->next_in = in_buf;
 340                         strm->avail_in = fread(in_buf, 1, BUFSIZ, file);
 341
 342                         if (ferror(file)) {
 343                                 // POSIX says that fread() sets errno if
 344                                 // an error occurred. ferror() doesn't
 345                                 // touch errno.
 346                                 my_errorf("%s: Error reading input file: %s",
 347                                                 filename, strerror(errno));
 348                                 exit(EXIT_FAILURE);
 349                         }
 350
 351 #ifndef LZMADEC
 352                         // When using LZMA_CONCATENATED, we need to tell
 353                         // liblzma when it has got all the input.
 354                         if (feof(file))
 355                                 action = LZMA_FINISH;
 356 #endif
 357                 }
 358
 359                 ret = lzma_code(strm, action);
 360
 361                 // Write and check write error before checking decoder error.
 362                 // This way as much data as possible gets written to output
 363                 // even if decoder detected an error.
 364                 if (strm->avail_out == 0 || ret != LZMA_OK) {
 365                         const size_t write_size = BUFSIZ - strm->avail_out;
 366
 367                         if (fwrite(out_buf, 1, write_size, stdout)
 368                                         != write_size) {
 369                                 // Wouldn't be a surprise if writing to stderr
 370                                 // would fail too but at least try to show an
 371                                 // error message.
 372                                 my_errorf("Cannot write to standard output: "
 373                                                 "%s", strerror(errno));
 374                                 exit(EXIT_FAILURE);
 375                         }
 376
 377                         strm->next_out = out_buf;
 378                         strm->avail_out = BUFSIZ;
 379                 }
 380
 381                 if (ret != LZMA_OK) {
 382                         if (ret == LZMA_STREAM_END) {
 383 #ifdef LZMADEC
 384                                 // Check that there's no trailing garbage.
 385                                 if (strm->avail_in != 0
 386                                                 || fread(in_buf, 1, 1, file)
 387                                                         != 0
 388                                                 || !feof(file))
 389                                         ret = LZMA_DATA_ERROR;
 390                                 else
 391                                         return;
 392 #else
 393                                 // lzma_stream_decoder() already guarantees
 394                                 // that there's no trailing garbage.
 395                                 assert(strm->avail_in == 0);
 396                                 assert(action == LZMA_FINISH);
 397                                 assert(feof(file));
 398                                 return;
 399 #endif
 400                         }
 401
 402                         const char *msg;
 403                         switch (ret) {
 404                         case LZMA_MEM_ERROR:
 405                                 msg = strerror(ENOMEM);
 406                                 break;
 407
 408                         case LZMA_MEMLIMIT_ERROR:
 409                                 msg = "Memory usage limit reached";
 410                                 break;
 411
 412                         case LZMA_FORMAT_ERROR:
 413                                 msg = "File format not recognized";
 414                                 break;
 415
 416                         case LZMA_OPTIONS_ERROR:
 417                                 // FIXME: Better message?
 418                                 msg = "Unsupported compression options";
 419                                 break;
 420
 421                         case LZMA_DATA_ERROR:
 422                                 msg = "File is corrupt";
 423                                 break;
 424
 425                         case LZMA_BUF_ERROR:
 426                                 msg = "Unexpected end of input";
 427                                 break;
 428
 429                         default:
 430                                 msg = "Internal error (bug)";
 431                                 break;
 432                         }
 433
 434                         my_errorf("%s: %s", filename, msg);
 435                         exit(EXIT_FAILURE);
 436                 }
 437         }
 438 }
 439
 440
 441 int
 442 main(int argc, char **argv)
 443 {
 444         // Set the argv0 global so that we can print the command name in
 445         // error and help messages.
 446         argv0 = argv[0];
 447
 448         // Set the default memory usage limit. This is needed before parsing
 449         // the command line arguments.
 450         memlimit_set(0);
 451
 452         // Parse the command line options.
 453         parse_options(argc, argv);
 454
 455         // The same lzma_stream is used for all files that we decode. This way
 456         // we don't need to reallocate memory for every file if they use same
 457         // compression settings.
 458         lzma_stream strm = LZMA_STREAM_INIT;
 459
 460         // Some systems require setting stdin and stdout to binary mode.
 461 #ifdef DOSLIKE
 462         setmode(fileno(stdin), O_BINARY);
 463         setmode(fileno(stdout), O_BINARY);
 464 #endif
 465
 466         if (optind == argc) {
 467                 // No filenames given, decode from stdin.
 468                 uncompress(&strm, stdin, "(stdin)");
 469         } else {
 470                 // Loop through the filenames given on the command line.
 471                 do {
 472                         // "-" indicates stdin.
 473                         if (strcmp(argv[optind], "-") == 0) {
 474                                 uncompress(&strm, stdin, "(stdin)");
 475                         } else {
 476                                 FILE *file = fopen(argv[optind], "rb");
 477                                 if (file == NULL) {
 478                                         my_errorf("%s: %s", argv[optind],
 479                                                         strerror(errno));
 480                                         exit(EXIT_FAILURE);
 481                                 }
 482
 483                                 uncompress(&strm, file, argv[optind]);
 484                                 fclose(file);
 485                         }
 486                 } while (++optind < argc);
 487         }
 488
 489 #ifndef NDEBUG
 490         // Free the memory only when debugging. Freeing wastes some time,
 491         // but allows detecting possible memory leaks with Valgrind.
 492         lzma_end(&strm);
 493 #endif
 494
 495         my_exit();
 496 }