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("Cannot write to standard output: %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 "  -f, --force        (ignored)\n"
  99 "  -M, --memory=NUM   use NUM bytes of memory at maximum (0 means default);\n"
 100 "                     the suffixes k, M, G, Ki, Mi, and Gi are supported.\n"
 101 "  -q, --quiet        specify *twice* to suppress errors\n"
 102 "  -Q, --no-warn      (ignored)\n"
 103 "  -h, --help         display this help and exit\n"
 104 "  -V, --version      display the version number and exit\n"
 105 "\n"
 106 "With no FILE, or when FILE is -, read standard input.\n"
 107 "\n"
 108 "On this system and configuration, this program will use at maximum of roughly\n"
 109 "%" PRIu64 " MiB RAM.\n"
 110 "\n"
 111 "Report bugs to <" PACKAGE_BUGREPORT "> (in English or Finnish).\n"
 112 "XZ Utils home page: <http://tukaani.org/xz/>\n",
 113                 argv0, memlimit / (1024 * 1024));
 114         my_exit();
 115 }
 116
 117
 118 static void lzma_attribute((noreturn))
 119 version(void)
 120 {
 121         printf(TOOL_FORMAT "dec " LZMA_VERSION_STRING "\n"
 122                         "liblzma %s\n", lzma_version_string());
 123
 124         my_exit();
 125 }
 126
 127
 128 /// Finds out the amount of physical memory in the system, and sets
 129 /// a default memory usage limit.
 130 static void
 131 set_default_memlimit(void)
 132 {
 133         const uint64_t mem = physmem();
 134
 135         if (mem == 0)
 136                 // Cannot autodetect, use 10 MiB as the default limit.
 137                 memlimit = UINT64_C(10) * 1024 * 1024;
 138         else
 139                 // Limit is 40 % of RAM.
 140                 memlimit = mem * 2 / 5;
 141
 142         return;
 143 }
 144
 145
 146 /// \brief      Converts a string to uint64_t
 147 ///
 148 /// This is rudely copied from src/xz/util.c and modified a little. :-(
 149 ///
 150 static uint64_t
 151 str_to_uint64(const char *value)
 152 {
 153         uint64_t result = 0;
 154
 155         // Accept special value "max".
 156         if (strcmp(value, "max") == 0)
 157                 return UINT64_MAX;
 158
 159         if (*value < '0' || *value > '9') {
 160                 my_errorf("%s: Not a number", value);
 161                 exit(EXIT_FAILURE);
 162         }
 163
 164         do {
 165                 // Don't overflow.
 166                 if (result > (UINT64_MAX - 9) / 10)
 167                         return UINT64_MAX;
 168
 169                 result *= 10;
 170                 result += *value - '0';
 171                 ++value;
 172         } while (*value >= '0' && *value <= '9');
 173
 174         if (*value != '\0') {
 175                 // Look for suffix.
 176                 static const struct {
 177                         const char name[4];
 178                         uint32_t multiplier;
 179                 } suffixes[] = {
 180                         { "k",   1000 },
 181                         { "kB",  1000 },
 182                         { "M",   1000000 },
 183                         { "MB",  1000000 },
 184                         { "G",   1000000000 },
 185                         { "GB",  1000000000 },
 186                         { "Ki",  1024 },
 187                         { "KiB", 1024 },
 188                         { "Mi",  1048576 },
 189                         { "MiB", 1048576 },
 190                         { "Gi",  1073741824 },
 191                         { "GiB", 1073741824 }
 192                 };
 193
 194                 uint32_t multiplier = 0;
 195                 for (size_t i = 0; i < ARRAY_SIZE(suffixes); ++i) {
 196                         if (strcmp(value, suffixes[i].name) == 0) {
 197                                 multiplier = suffixes[i].multiplier;
 198                                 break;
 199                         }
 200                 }
 201
 202                 if (multiplier == 0) {
 203                         my_errorf("%s: Invalid suffix", value);
 204                         exit(EXIT_FAILURE);
 205                 }
 206
 207                 // Don't overflow here either.
 208                 if (result > UINT64_MAX / multiplier)
 209                         result = UINT64_MAX;
 210                 else
 211                         result *= multiplier;
 212         }
 213
 214         return result;
 215 }
 216
 217
 218 /// Parses command line options.
 219 static void
 220 parse_options(int argc, char **argv)
 221 {
 222         static const char short_opts[] = "cdkfM:hqQV";
 223         static const struct option long_opts[] = {
 224                 { "stdout",       no_argument,         NULL, 'c' },
 225                 { "to-stdout",    no_argument,         NULL, 'c' },
 226                 { "decompress",   no_argument,         NULL, 'd' },
 227                 { "uncompress",   no_argument,         NULL, 'd' },
 228                 { "force",        no_argument,         NULL, 'f' },
 229                 { "keep",         no_argument,         NULL, 'k' },
 230                 { "memory",       required_argument,   NULL, 'M' },
 231                 { "quiet",        no_argument,         NULL, 'q' },
 232                 { "no-warn",      no_argument,         NULL, 'Q' },
 233                 { "help",         no_argument,         NULL, 'h' },
 234                 { "version",      no_argument,         NULL, 'V' },
 235                 { NULL,           0,                   NULL, 0   }
 236         };
 237
 238         int c;
 239
 240         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 241                         != -1) {
 242                 switch (c) {
 243                 case 'c':
 244                 case 'd':
 245                 case 'f':
 246                 case 'k':
 247                 case 'Q':
 248                         break;
 249
 250                 case 'M':
 251                         memlimit = str_to_uint64(optarg);
 252                         if (memlimit == 0)
 253                                 set_default_memlimit();
 254
 255                         break;
 256
 257                 case 'q':
 258                         if (display_errors > 0)
 259                                 --display_errors;
 260
 261                         break;
 262
 263                 case 'h':
 264                         help();
 265
 266                 case 'V':
 267                         version();
 268
 269                 default:
 270                         exit(EXIT_FAILURE);
 271                 }
 272         }
 273
 274         return;
 275 }
 276
 277
 278 static void
 279 uncompress(lzma_stream *strm, FILE *file, const char *filename)
 280 {
 281         lzma_ret ret;
 282
 283         // Initialize the decoder
 284 #ifdef LZMADEC
 285         ret = lzma_alone_decoder(strm, memlimit);
 286 #else
 287         ret = lzma_stream_decoder(strm, memlimit, LZMA_CONCATENATED);
 288 #endif
 289
 290         // The only reasonable error here is LZMA_MEM_ERROR.
 291         // FIXME: Maybe also LZMA_MEMLIMIT_ERROR in future?
 292         if (ret != LZMA_OK) {
 293                 my_errorf("%s", ret == LZMA_MEM_ERROR ? strerror(ENOMEM)
 294                                 : "Internal program error (bug)");
 295                 exit(EXIT_FAILURE);
 296         }
 297
 298         // Input and output buffers
 299         uint8_t in_buf[BUFSIZ];
 300         uint8_t out_buf[BUFSIZ];
 301
 302         strm->avail_in = 0;
 303         strm->next_out = out_buf;
 304         strm->avail_out = BUFSIZ;
 305
 306         lzma_action action = LZMA_RUN;
 307
 308         while (true) {
 309                 if (strm->avail_in == 0) {
 310                         strm->next_in = in_buf;
 311                         strm->avail_in = fread(in_buf, 1, BUFSIZ, file);
 312
 313                         if (ferror(file)) {
 314                                 // POSIX says that fread() sets errno if
 315                                 // an error occurred. ferror() doesn't
 316                                 // touch errno.
 317                                 my_errorf("%s: Error reading input file: %s",
 318                                                 filename, strerror(errno));
 319                                 exit(EXIT_FAILURE);
 320                         }
 321
 322 #ifndef LZMADEC
 323                         // When using LZMA_CONCATENATED, we need to tell
 324                         // liblzma when it has got all the input.
 325                         if (feof(file))
 326                                 action = LZMA_FINISH;
 327 #endif
 328                 }
 329
 330                 ret = lzma_code(strm, action);
 331
 332                 // Write and check write error before checking decoder error.
 333                 // This way as much data as possible gets written to output
 334                 // even if decoder detected an error.
 335                 if (strm->avail_out == 0 || ret != LZMA_OK) {
 336                         const size_t write_size = BUFSIZ - strm->avail_out;
 337
 338                         if (fwrite(out_buf, 1, write_size, stdout)
 339                                         != write_size) {
 340                                 // Wouldn't be a surprise if writing to stderr
 341                                 // would fail too but at least try to show an
 342                                 // error message.
 343                                 my_errorf("Cannot write to standard output: "
 344                                                 "%s", strerror(errno));
 345                                 exit(EXIT_FAILURE);
 346                         }
 347
 348                         strm->next_out = out_buf;
 349                         strm->avail_out = BUFSIZ;
 350                 }
 351
 352                 if (ret != LZMA_OK) {
 353                         if (ret == LZMA_STREAM_END) {
 354 #ifdef LZMADEC
 355                                 // Check that there's no trailing garbage.
 356                                 if (strm->avail_in != 0
 357                                                 || fread(in_buf, 1, 1, file)
 358                                                         != 0
 359                                                 || !feof(file))
 360                                         ret = LZMA_DATA_ERROR;
 361                                 else
 362                                         return;
 363 #else
 364                                 // lzma_stream_decoder() already guarantees
 365                                 // that there's no trailing garbage.
 366                                 assert(strm->avail_in == 0);
 367                                 assert(action == LZMA_FINISH);
 368                                 assert(feof(file));
 369                                 return;
 370 #endif
 371                         }
 372
 373                         const char *msg;
 374                         switch (ret) {
 375                         case LZMA_MEM_ERROR:
 376                                 msg = strerror(ENOMEM);
 377                                 break;
 378
 379                         case LZMA_MEMLIMIT_ERROR:
 380                                 msg = "Memory usage limit reached";
 381                                 break;
 382
 383                         case LZMA_FORMAT_ERROR:
 384                                 msg = "File format not recognized";
 385                                 break;
 386
 387                         case LZMA_OPTIONS_ERROR:
 388                                 // FIXME: Better message?
 389                                 msg = "Unsupported compression options";
 390                                 break;
 391
 392                         case LZMA_DATA_ERROR:
 393                                 msg = "File is corrupt";
 394                                 break;
 395
 396                         case LZMA_BUF_ERROR:
 397                                 msg = "Unexpected end of input";
 398                                 break;
 399
 400                         default:
 401                                 msg = "Internal program error (bug)";
 402                                 break;
 403                         }
 404
 405                         my_errorf("%s: %s", filename, msg);
 406                         exit(EXIT_FAILURE);
 407                 }
 408         }
 409 }
 410
 411
 412 int
 413 main(int argc, char **argv)
 414 {
 415         // Set the argv0 global so that we can print the command name in
 416         // error and help messages.
 417         argv0 = argv[0];
 418
 419         // Detect amount of installed RAM and set the memory usage limit.
 420         // This is needed before parsing the command line arguments.
 421         set_default_memlimit();
 422
 423         // Parse the command line options.
 424         parse_options(argc, argv);
 425
 426         // The same lzma_stream is used for all files that we decode. This way
 427         // we don't need to reallocate memory for every file if they use same
 428         // compression settings.
 429         lzma_stream strm = LZMA_STREAM_INIT;
 430
 431         // Some systems require setting stdin and stdout to binary mode.
 432 #ifdef DOSLIKE
 433         setmode(fileno(stdin), O_BINARY);
 434         setmode(fileno(stdout), O_BINARY);
 435 #endif
 436
 437         if (optind == argc) {
 438                 // No filenames given, decode from stdin.
 439                 uncompress(&strm, stdin, "(stdin)");
 440         } else {
 441                 // Loop through the filenames given on the command line.
 442                 do {
 443                         // "-" indicates stdin.
 444                         if (strcmp(argv[optind], "-") == 0) {
 445                                 uncompress(&strm, stdin, "(stdin)");
 446                         } else {
 447                                 FILE *file = fopen(argv[optind], "rb");
 448                                 if (file == NULL) {
 449                                         my_errorf("%s: %s", argv[optind],
 450                                                         strerror(errno));
 451                                         exit(EXIT_FAILURE);
 452                                 }
 453
 454                                 uncompress(&strm, file, argv[optind]);
 455                                 fclose(file);
 456                         }
 457                 } while (++optind < argc);
 458         }
 459
 460 #ifndef NDEBUG
 461         // Free the memory only when debugging. Freeing wastes some time,
 462         // but allows detecting possible memory leaks with Valgrind.
 463         lzma_end(&strm);
 464 #endif
 465
 466         my_exit();
 467 }