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