src/xz/args.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       args.c
   4 /// \brief      Argument parsing
   5 ///
   6 /// \note       Filter-specific options parsing is in options.c.
   7 //
   8 //  Author:     Lasse Collin
   9 //
  10 //  This file has been put into the public domain.
  11 //  You can do whatever you want with this file.
  12 //
  13 ///////////////////////////////////////////////////////////////////////////////
  14
  15 #include "private.h"
  16
  17 #include "getopt.h"
  18 #include <ctype.h>
  19
  20
  21 bool opt_stdout = false;
  22 bool opt_force = false;
  23 bool opt_keep_original = false;
  24 bool opt_robot = false;
  25
  26 // We don't modify or free() this, but we need to assign it in some
  27 // non-const pointers.
  28 const char *stdin_filename = "(stdin)";
  29
  30
  31 static void
  32 parse_real(args_info *args, int argc, char **argv)
  33 {
  34         enum {
  35                 OPT_X86 = INT_MIN,
  36                 OPT_POWERPC,
  37                 OPT_IA64,
  38                 OPT_ARM,
  39                 OPT_ARMTHUMB,
  40                 OPT_SPARC,
  41                 OPT_DELTA,
  42                 OPT_LZMA1,
  43                 OPT_LZMA2,
  44
  45                 OPT_NO_SPARSE,
  46                 OPT_FILES,
  47                 OPT_FILES0,
  48                 OPT_NO_ADJUST,
  49                 OPT_INFO_MEMORY,
  50                 OPT_ROBOT,
  51         };
  52
  53         static const char short_opts[]
  54                         = "cC:defF:hHlkM:qQrS:tT:vVz0123456789";
  55
  56         static const struct option long_opts[] = {
  57                 // Operation mode
  58                 { "compress",     no_argument,       NULL,  'z' },
  59                 { "decompress",   no_argument,       NULL,  'd' },
  60                 { "uncompress",   no_argument,       NULL,  'd' },
  61                 { "test",         no_argument,       NULL,  't' },
  62                 { "list",         no_argument,       NULL,  'l' },
  63
  64                 // Operation modifiers
  65                 { "keep",         no_argument,       NULL,  'k' },
  66                 { "force",        no_argument,       NULL,  'f' },
  67                 { "stdout",       no_argument,       NULL,  'c' },
  68                 { "to-stdout",    no_argument,       NULL,  'c' },
  69                 { "no-sparse",    no_argument,       NULL,  OPT_NO_SPARSE },
  70                 { "suffix",       required_argument, NULL,  'S' },
  71                 // { "recursive",      no_argument,       NULL,  'r' }, // TODO
  72                 { "files",        optional_argument, NULL,  OPT_FILES },
  73                 { "files0",       optional_argument, NULL,  OPT_FILES0 },
  74
  75                 // Basic compression settings
  76                 { "format",       required_argument, NULL,  'F' },
  77                 { "check",        required_argument, NULL,  'C' },
  78                 { "no-adjust",    no_argument,       NULL,  OPT_NO_ADJUST },
  79                 { "memory",       required_argument, NULL,  'M' },
  80                 { "threads",      required_argument, NULL,  'T' },
  81
  82                 { "extreme",      no_argument,       NULL,  'e' },
  83                 { "fast",         no_argument,       NULL,  '0' },
  84                 { "best",         no_argument,       NULL,  '9' },
  85
  86                 // Filters
  87                 { "lzma1",        optional_argument, NULL,  OPT_LZMA1 },
  88                 { "lzma2",        optional_argument, NULL,  OPT_LZMA2 },
  89                 { "x86",          optional_argument, NULL,  OPT_X86 },
  90                 { "powerpc",      optional_argument, NULL,  OPT_POWERPC },
  91                 { "ia64",         optional_argument, NULL,  OPT_IA64 },
  92                 { "arm",          optional_argument, NULL,  OPT_ARM },
  93                 { "armthumb",     optional_argument, NULL,  OPT_ARMTHUMB },
  94                 { "sparc",        optional_argument, NULL,  OPT_SPARC },
  95                 { "delta",        optional_argument, NULL,  OPT_DELTA },
  96
  97                 // Other options
  98                 { "quiet",        no_argument,       NULL,  'q' },
  99                 { "verbose",      no_argument,       NULL,  'v' },
 100                 { "no-warn",      no_argument,       NULL,  'Q' },
 101                 { "robot",        no_argument,       NULL,  OPT_ROBOT },
 102                 { "info-memory",  no_argument,       NULL,  OPT_INFO_MEMORY },
 103                 { "help",         no_argument,       NULL,  'h' },
 104                 { "long-help",    no_argument,       NULL,  'H' },
 105                 { "version",      no_argument,       NULL,  'V' },
 106
 107                 { NULL,                 0,                 NULL,   0 }
 108         };
 109
 110         int c;
 111
 112         while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
 113                         != -1) {
 114                 switch (c) {
 115                 // Compression preset (also for decompression if --format=raw)
 116                 case '0': case '1': case '2': case '3': case '4':
 117                 case '5': case '6': case '7': case '8': case '9':
 118                         coder_set_preset(c - '0');
 119                         break;
 120
 121                 // --memory
 122                 case 'M': {
 123                         // Support specifying the limit as a percentage of
 124                         // installed physical RAM.
 125                         size_t len = strlen(optarg);
 126                         if (len > 0 && optarg[len - 1] == '%') {
 127                                 optarg[len - 1] = '\0';
 128                                 hardware_memlimit_set_percentage(
 129                                                 str_to_uint64(
 130                                                 "memory%", optarg, 1, 100));
 131                         } else {
 132                                 // On 32-bit systems, SIZE_MAX would make more
 133                                 // sense than UINT64_MAX. But use UINT64_MAX
 134                                 // still so that scripts that assume > 4 GiB
 135                                 // values don't break.
 136                                 hardware_memlimit_set(str_to_uint64(
 137                                                 "memory", optarg,
 138                                                 0, UINT64_MAX));
 139                         }
 140
 141                         break;
 142                 }
 143
 144                 // --suffix
 145                 case 'S':
 146                         suffix_set(optarg);
 147                         break;
 148
 149                 case 'T':
 150                         hardware_threadlimit_set(str_to_uint64(
 151                                         "threads", optarg, 0, UINT32_MAX));
 152                         break;
 153
 154                 // --version
 155                 case 'V':
 156                         // This doesn't return.
 157                         message_version();
 158
 159                 // --stdout
 160                 case 'c':
 161                         opt_stdout = true;
 162                         break;
 163
 164                 // --decompress
 165                 case 'd':
 166                         opt_mode = MODE_DECOMPRESS;
 167                         break;
 168
 169                 // --extreme
 170                 case 'e':
 171                         coder_set_extreme();
 172                         break;
 173
 174                 // --force
 175                 case 'f':
 176                         opt_force = true;
 177                         break;
 178
 179                 // --info-memory
 180                 case OPT_INFO_MEMORY:
 181                         // This doesn't return.
 182                         message_memlimit();
 183
 184                 // --help
 185                 case 'h':
 186                         // This doesn't return.
 187                         message_help(false);
 188
 189                 // --long-help
 190                 case 'H':
 191                         // This doesn't return.
 192                         message_help(true);
 193
 194                 // --list
 195                 case 'l':
 196                         opt_mode = MODE_LIST;
 197                         break;
 198
 199                 // --keep
 200                 case 'k':
 201                         opt_keep_original = true;
 202                         break;
 203
 204                 // --quiet
 205                 case 'q':
 206                         message_verbosity_decrease();
 207                         break;
 208
 209                 case 'Q':
 210                         set_exit_no_warn();
 211                         break;
 212
 213                 case 't':
 214                         opt_mode = MODE_TEST;
 215                         break;
 216
 217                 // --verbose
 218                 case 'v':
 219                         message_verbosity_increase();
 220                         break;
 221
 222                 // --robot
 223                 case OPT_ROBOT:
 224                         opt_robot = true;
 225
 226                         // This is to make sure that floating point numbers
 227                         // always have a dot as decimal separator.
 228                         setlocale(LC_NUMERIC, "C");
 229                         break;
 230
 231                 case 'z':
 232                         opt_mode = MODE_COMPRESS;
 233                         break;
 234
 235                 // Filter setup
 236
 237                 case OPT_X86:
 238                         coder_add_filter(LZMA_FILTER_X86,
 239                                         options_bcj(optarg));
 240                         break;
 241
 242                 case OPT_POWERPC:
 243                         coder_add_filter(LZMA_FILTER_POWERPC,
 244                                         options_bcj(optarg));
 245                         break;
 246
 247                 case OPT_IA64:
 248                         coder_add_filter(LZMA_FILTER_IA64,
 249                                         options_bcj(optarg));
 250                         break;
 251
 252                 case OPT_ARM:
 253                         coder_add_filter(LZMA_FILTER_ARM,
 254                                         options_bcj(optarg));
 255                         break;
 256
 257                 case OPT_ARMTHUMB:
 258                         coder_add_filter(LZMA_FILTER_ARMTHUMB,
 259                                         options_bcj(optarg));
 260                         break;
 261
 262                 case OPT_SPARC:
 263                         coder_add_filter(LZMA_FILTER_SPARC,
 264                                         options_bcj(optarg));
 265                         break;
 266
 267                 case OPT_DELTA:
 268                         coder_add_filter(LZMA_FILTER_DELTA,
 269                                         options_delta(optarg));
 270                         break;
 271
 272                 case OPT_LZMA1:
 273                         coder_add_filter(LZMA_FILTER_LZMA1,
 274                                         options_lzma(optarg));
 275                         break;
 276
 277                 case OPT_LZMA2:
 278                         coder_add_filter(LZMA_FILTER_LZMA2,
 279                                         options_lzma(optarg));
 280                         break;
 281
 282                 // Other
 283
 284                 // --format
 285                 case 'F': {
 286                         // Just in case, support both "lzma" and "alone" since
 287                         // the latter was used for forward compatibility in
 288                         // LZMA Utils 4.32.x.
 289                         static const struct {
 290                                 char str[8];
 291                                 enum format_type format;
 292                         } types[] = {
 293                                 { "auto",   FORMAT_AUTO },
 294                                 { "xz",     FORMAT_XZ },
 295                                 { "lzma",   FORMAT_LZMA },
 296                                 { "alone",  FORMAT_LZMA },
 297                                 // { "gzip",   FORMAT_GZIP },
 298                                 // { "gz",     FORMAT_GZIP },
 299                                 { "raw",    FORMAT_RAW },
 300                         };
 301
 302                         size_t i = 0;
 303                         while (strcmp(types[i].str, optarg) != 0)
 304                                 if (++i == ARRAY_SIZE(types))
 305                                         message_fatal(_("%s: Unknown file "
 306                                                         "format type"),
 307                                                         optarg);
 308
 309                         opt_format = types[i].format;
 310                         break;
 311                 }
 312
 313                 // --check
 314                 case 'C': {
 315                         static const struct {
 316                                 char str[8];
 317                                 lzma_check check;
 318                         } types[] = {
 319                                 { "none",   LZMA_CHECK_NONE },
 320                                 { "crc32",  LZMA_CHECK_CRC32 },
 321                                 { "crc64",  LZMA_CHECK_CRC64 },
 322                                 { "sha256", LZMA_CHECK_SHA256 },
 323                         };
 324
 325                         size_t i = 0;
 326                         while (strcmp(types[i].str, optarg) != 0) {
 327                                 if (++i == ARRAY_SIZE(types))
 328                                         message_fatal(_("%s: Unsupported "
 329                                                         "integrity "
 330                                                         "check type"), optarg);
 331                         }
 332
 333                         // Use a separate check in case we are using different
 334                         // liblzma than what was used to compile us.
 335                         if (!lzma_check_is_supported(types[i].check))
 336                                 message_fatal(_("%s: Unsupported integrity "
 337                                                 "check type"), optarg);
 338
 339                         coder_set_check(types[i].check);
 340                         break;
 341                 }
 342
 343                 case OPT_NO_SPARSE:
 344                         io_no_sparse();
 345                         break;
 346
 347                 case OPT_FILES:
 348                         args->files_delim = '\n';
 349
 350                 // Fall through
 351
 352                 case OPT_FILES0:
 353                         if (args->files_name != NULL)
 354                                 message_fatal(_("Only one file can be "
 355                                                 "specified with `--files' "
 356                                                 "or `--files0'."));
 357
 358                         if (optarg == NULL) {
 359                                 args->files_name = (char *)stdin_filename;
 360                                 args->files_file = stdin;
 361                         } else {
 362                                 args->files_name = optarg;
 363                                 args->files_file = fopen(optarg,
 364                                                 c == OPT_FILES ? "r" : "rb");
 365                                 if (args->files_file == NULL)
 366                                         message_fatal("%s: %s", optarg,
 367                                                         strerror(errno));
 368                         }
 369
 370                         break;
 371
 372                 case OPT_NO_ADJUST:
 373                         opt_auto_adjust = false;
 374                         break;
 375
 376                 default:
 377                         message_try_help();
 378                         tuklib_exit(E_ERROR, E_ERROR, false);
 379                 }
 380         }
 381
 382         return;
 383 }
 384
 385
 386 static void
 387 parse_environment(args_info *args, char *argv0)
 388 {
 389         char *env = getenv("XZ_OPT");
 390         if (env == NULL)
 391                 return;
 392
 393         // We modify the string, so make a copy of it.
 394         env = xstrdup(env);
 395
 396         // Calculate the number of arguments in env. argc stats at one
 397         // to include space for the program name.
 398         int argc = 1;
 399         bool prev_was_space = true;
 400         for (size_t i = 0; env[i] != '\0'; ++i) {
 401                 // NOTE: Cast to unsigned char is needed so that correct
 402                 // value gets passed to isspace(), which expects
 403                 // unsigned char cast to int. Casting to int is done
 404                 // automatically due to integer promotion, but we need to
 405                 // force char to unsigned char manually. Otherwise 8-bit
 406                 // characters would get promoted to wrong value if
 407                 // char is signed.
 408                 if (isspace((unsigned char)env[i])) {
 409                         prev_was_space = true;
 410                 } else if (prev_was_space) {
 411                         prev_was_space = false;
 412
 413                         // Keep argc small enough to fit into a singed int
 414                         // and to keep it usable for memory allocation.
 415                         if (++argc == my_min(
 416                                         INT_MAX, SIZE_MAX / sizeof(char *)))
 417                                 message_fatal(_("The environment variable "
 418                                                 "XZ_OPT contains too many "
 419                                                 "arguments"));
 420                 }
 421         }
 422
 423         // Allocate memory to hold pointers to the arguments. Add one to get
 424         // space for the terminating NULL (if some systems happen to need it).
 425         char **argv = xmalloc(((size_t)(argc) + 1) * sizeof(char *));
 426         argv[0] = argv0;
 427         argv[argc] = NULL;
 428
 429         // Go through the string again. Split the arguments using '\0'
 430         // characters and add pointers to the resulting strings to argv.
 431         argc = 1;
 432         prev_was_space = true;
 433         for (size_t i = 0; env[i] != '\0'; ++i) {
 434                 if (isspace((unsigned char)env[i])) {
 435                         prev_was_space = true;
 436                         env[i] = '\0';
 437                 } else if (prev_was_space) {
 438                         prev_was_space = false;
 439                         argv[argc++] = env + i;
 440                 }
 441         }
 442
 443         // Parse the argument list we got from the environment. All non-option
 444         // arguments i.e. filenames are ignored.
 445         parse_real(args, argc, argv);
 446
 447         // Reset the state of the getopt_long() so that we can parse the
 448         // command line options too. There are two incompatible ways to
 449         // do it.
 450 #ifdef HAVE_OPTRESET
 451         // BSD
 452         optind = 1;
 453         optreset = 1;
 454 #else
 455         // GNU, Solaris
 456         optind = 0;
 457 #endif
 458
 459         // We don't need the argument list from environment anymore.
 460         free(argv);
 461         free(env);
 462
 463         return;
 464 }
 465
 466
 467 extern void
 468 args_parse(args_info *args, int argc, char **argv)
 469 {
 470         // Initialize those parts of *args that we need later.
 471         args->files_name = NULL;
 472         args->files_file = NULL;
 473         args->files_delim = '\0';
 474
 475         // Check how we were called.
 476         {
 477                 // Remove the leading path name, if any.
 478                 const char *name = strrchr(argv[0], '/');
 479                 if (name == NULL)
 480                         name = argv[0];
 481                 else
 482                         ++name;
 483
 484                 // NOTE: It's possible that name[0] is now '\0' if argv[0]
 485                 // is weird, but it doesn't matter here.
 486
 487                 // Look for full command names instead of substrings like
 488                 // "un", "cat", and "lz" to reduce possibility of false
 489                 // positives when the programs have been renamed.
 490                 if (strstr(name, "xzcat") != NULL) {
 491                         opt_mode = MODE_DECOMPRESS;
 492                         opt_stdout = true;
 493                 } else if (strstr(name, "unxz") != NULL) {
 494                         opt_mode = MODE_DECOMPRESS;
 495                 } else if (strstr(name, "lzcat") != NULL) {
 496                         opt_format = FORMAT_LZMA;
 497                         opt_mode = MODE_DECOMPRESS;
 498                         opt_stdout = true;
 499                 } else if (strstr(name, "unlzma") != NULL) {
 500                         opt_format = FORMAT_LZMA;
 501                         opt_mode = MODE_DECOMPRESS;
 502                 } else if (strstr(name, "lzma") != NULL) {
 503                         opt_format = FORMAT_LZMA;
 504                 }
 505         }
 506
 507         // First the flags from environment
 508         parse_environment(args, argv[0]);
 509
 510         // Then from the command line
 511         parse_real(args, argc, argv);
 512
 513         // Never remove the source file when the destination is not on disk.
 514         // In test mode the data is written nowhere, but setting opt_stdout
 515         // will make the rest of the code behave well.
 516         if (opt_stdout || opt_mode == MODE_TEST) {
 517                 opt_keep_original = true;
 518                 opt_stdout = true;
 519         }
 520
 521         // When compressing, if no --format flag was used, or it
 522         // was --format=auto, we compress to the .xz format.
 523         if (opt_mode == MODE_COMPRESS && opt_format == FORMAT_AUTO)
 524                 opt_format = FORMAT_XZ;
 525
 526         // Compression settings need to be validated (options themselves and
 527         // their memory usage) when compressing to any file format. It has to
 528         // be done also when uncompressing raw data, since for raw decoding
 529         // the options given on the command line are used to know what kind
 530         // of raw data we are supposed to decode.
 531         if (opt_mode == MODE_COMPRESS || opt_format == FORMAT_RAW)
 532                 coder_set_compression_settings();
 533
 534         // If no filenames are given, use stdin.
 535         if (argv[optind] == NULL && args->files_name == NULL) {
 536                 // We don't modify or free() the "-" constant. The caller
 537                 // modifies this so don't make the struct itself const.
 538                 static char *names_stdin[2] = { (char *)"-", NULL };
 539                 args->arg_names = names_stdin;
 540                 args->arg_count = 1;
 541         } else {
 542                 // We got at least one filename from the command line, or
 543                 // --files or --files0 was specified.
 544                 args->arg_names = argv + optind;
 545                 args->arg_count = argc - optind;
 546         }
 547
 548         return;
 549 }