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